2024-01-16 09:53:51,486 - INFO - root - Hello! This is Joey-NMT (version 2.3.0). 2024-01-16 09:53:51,602 - INFO - root - Spawn torch.multiprocessing (nprocs=2). 2024-01-16 09:53:55,380 - INFO - joeynmt.config - cfg.name : iwslt14_deenfr_prompt 2024-01-16 09:53:55,380 - INFO - joeynmt.config - cfg.joeynmt_version : 2.3.0 2024-01-16 09:53:55,380 - INFO - joeynmt.config - cfg.model_dir : models/iwslt14_prompt 2024-01-16 09:53:55,380 - INFO - joeynmt.config - cfg.use_cuda : True 2024-01-16 09:53:55,381 - INFO - joeynmt.config - cfg.fp16 : True 2024-01-16 09:53:55,381 - INFO - joeynmt.config - cfg.random_seed : 42 2024-01-16 09:53:55,381 - INFO - joeynmt.config - cfg.data.train : test/data/iwslt14_prompt/train 2024-01-16 09:53:55,381 - INFO - joeynmt.config - cfg.data.dev : test/data/iwslt14_prompt/dev 2024-01-16 09:53:55,381 - INFO - joeynmt.config - cfg.data.test : test/data/iwslt14_prompt/test 2024-01-16 09:53:55,381 - INFO - joeynmt.config - cfg.data.dataset_type : tsv 2024-01-16 09:53:55,381 - INFO - joeynmt.config - cfg.data.sample_dev_subset : 500 2024-01-16 09:53:55,381 - INFO - joeynmt.config - cfg.data.src.lang : src 2024-01-16 09:53:55,381 - INFO - joeynmt.config - cfg.data.src.max_length : 512 2024-01-16 09:53:55,381 - INFO - joeynmt.config - cfg.data.src.lowercase : False 2024-01-16 09:53:55,381 - INFO - joeynmt.config - cfg.data.src.normalize : False 2024-01-16 09:53:55,381 - INFO - joeynmt.config - cfg.data.src.level : bpe 2024-01-16 09:53:55,381 - INFO - joeynmt.config - cfg.data.src.voc_limit : 32000 2024-01-16 09:53:55,381 - INFO - joeynmt.config - cfg.data.src.voc_min_freq : 1 2024-01-16 09:53:55,381 - INFO - joeynmt.config - cfg.data.src.voc_file : test/data/iwslt14_prompt/sp.vocab 2024-01-16 09:53:55,381 - INFO - joeynmt.config - cfg.data.src.tokenizer_type : sentencepiece 2024-01-16 09:53:55,381 - INFO - joeynmt.config - cfg.data.src.tokenizer_cfg.model_file : test/data/iwslt14_prompt/sp.model 2024-01-16 09:53:55,381 - INFO - joeynmt.config - cfg.data.src.tokenizer_cfg.model_type : unigram 2024-01-16 09:53:55,381 - INFO - joeynmt.config - cfg.data.src.tokenizer_cfg.character_coverage : 1.0 2024-01-16 09:53:55,381 - INFO - joeynmt.config - cfg.data.trg.lang : trg 2024-01-16 09:53:55,381 - INFO - joeynmt.config - cfg.data.trg.max_length : 512 2024-01-16 09:53:55,381 - INFO - joeynmt.config - cfg.data.trg.lowercase : False 2024-01-16 09:53:55,381 - INFO - joeynmt.config - cfg.data.trg.normalize : False 2024-01-16 09:53:55,382 - INFO - joeynmt.config - cfg.data.trg.level : bpe 2024-01-16 09:53:55,382 - INFO - joeynmt.config - cfg.data.trg.voc_limit : 32000 2024-01-16 09:53:55,382 - INFO - joeynmt.config - cfg.data.trg.voc_min_freq : 1 2024-01-16 09:53:55,382 - INFO - joeynmt.config - cfg.data.trg.voc_file : test/data/iwslt14_prompt/sp.vocab 2024-01-16 09:53:55,382 - INFO - joeynmt.config - cfg.data.trg.tokenizer_type : sentencepiece 2024-01-16 09:53:55,382 - INFO - joeynmt.config - cfg.data.trg.tokenizer_cfg.model_file : test/data/iwslt14_prompt/sp.model 2024-01-16 09:53:55,382 - INFO - joeynmt.config - cfg.data.trg.tokenizer_cfg.model_type : unigram 2024-01-16 09:53:55,382 - INFO - joeynmt.config - cfg.data.trg.tokenizer_cfg.character_coverage : 1.0 2024-01-16 09:53:55,382 - INFO - joeynmt.config - cfg.data.special_symbols.unk_token : 2024-01-16 09:53:55,382 - INFO - joeynmt.config - cfg.data.special_symbols.unk_id : 0 2024-01-16 09:53:55,382 - INFO - joeynmt.config - cfg.data.special_symbols.pad_token : 2024-01-16 09:53:55,382 - INFO - joeynmt.config - cfg.data.special_symbols.pad_id : 1 2024-01-16 09:53:55,382 - INFO - joeynmt.config - cfg.data.special_symbols.bos_token : 2024-01-16 09:53:55,382 - INFO - joeynmt.config - cfg.data.special_symbols.bos_id : 2 2024-01-16 09:53:55,382 - INFO - joeynmt.config - cfg.data.special_symbols.eos_token : 2024-01-16 09:53:55,382 - INFO - joeynmt.config - cfg.data.special_symbols.eos_id : 3 2024-01-16 09:53:55,382 - INFO - joeynmt.config - cfg.data.special_symbols.sep_token : 2024-01-16 09:53:55,382 - INFO - joeynmt.config - cfg.data.special_symbols.sep_id : 4 2024-01-16 09:53:55,382 - INFO - joeynmt.config - cfg.data.special_symbols.lang_tags : ['', '', ''] 2024-01-16 09:53:55,382 - INFO - joeynmt.config - cfg.testing.n_best : 1 2024-01-16 09:53:55,382 - INFO - joeynmt.config - cfg.testing.beam_size : 5 2024-01-16 09:53:55,382 - INFO - joeynmt.config - cfg.testing.beam_alpha : 1.0 2024-01-16 09:53:55,382 - INFO - joeynmt.config - cfg.testing.batch_size : 32 2024-01-16 09:53:55,382 - INFO - joeynmt.config - cfg.testing.batch_type : sentence 2024-01-16 09:53:55,382 - INFO - joeynmt.config - cfg.testing.max_output_length : 512 2024-01-16 09:53:55,383 - INFO - joeynmt.config - cfg.testing.eval_metrics : ['bleu'] 2024-01-16 09:53:55,383 - INFO - joeynmt.config - cfg.testing.sacrebleu_cfg.tokenize : 13a 2024-01-16 09:53:55,383 - INFO - joeynmt.config - cfg.testing.sacrebleu_cfg.lowercase : True 2024-01-16 09:53:55,383 - INFO - joeynmt.config - cfg.training.optimizer : adamw 2024-01-16 09:53:55,383 - INFO - joeynmt.config - cfg.training.normalization : tokens 2024-01-16 09:53:55,383 - INFO - joeynmt.config - cfg.training.adam_betas : [0.9, 0.98] 2024-01-16 09:53:55,383 - INFO - joeynmt.config - cfg.training.scheduling : warmupinversesquareroot 2024-01-16 09:53:55,383 - INFO - joeynmt.config - cfg.training.learning_rate_warmup : 10000 2024-01-16 09:53:55,383 - INFO - joeynmt.config - cfg.training.learning_rate : 0.0002 2024-01-16 09:53:55,383 - INFO - joeynmt.config - cfg.training.learning_rate_min : 1e-07 2024-01-16 09:53:55,383 - INFO - joeynmt.config - cfg.training.weight_decay : 0.001 2024-01-16 09:53:55,383 - INFO - joeynmt.config - cfg.training.label_smoothing : 0.1 2024-01-16 09:53:55,383 - INFO - joeynmt.config - cfg.training.loss : crossentropy 2024-01-16 09:53:55,383 - INFO - joeynmt.config - cfg.training.batch_size : 32 2024-01-16 09:53:55,383 - INFO - joeynmt.config - cfg.training.batch_type : sentence 2024-01-16 09:53:55,383 - INFO - joeynmt.config - cfg.training.batch_multiplier : 4 2024-01-16 09:53:55,383 - INFO - joeynmt.config - cfg.training.early_stopping_metric : bleu 2024-01-16 09:53:55,383 - INFO - joeynmt.config - cfg.training.epochs : 50 2024-01-16 09:53:55,383 - INFO - joeynmt.config - cfg.training.validation_freq : 1000 2024-01-16 09:53:55,383 - INFO - joeynmt.config - cfg.training.logging_freq : 100 2024-01-16 09:53:55,383 - INFO - joeynmt.config - cfg.training.overwrite : False 2024-01-16 09:53:55,383 - INFO - joeynmt.config - cfg.training.shuffle : True 2024-01-16 09:53:55,383 - INFO - joeynmt.config - cfg.training.print_valid_sents : [0, 1, 2, 3] 2024-01-16 09:53:55,383 - INFO - joeynmt.config - cfg.training.keep_best_ckpts : 5 2024-01-16 09:53:55,383 - INFO - joeynmt.config - cfg.model.initializer : xavier_uniform 2024-01-16 09:53:55,384 - INFO - joeynmt.config - cfg.model.bias_initializer : zeros 2024-01-16 09:53:55,384 - INFO - joeynmt.config - cfg.model.init_gain : 1.0 2024-01-16 09:53:55,384 - INFO - joeynmt.config - cfg.model.embed_initializer : xavier_uniform 2024-01-16 09:53:55,384 - INFO - joeynmt.config - cfg.model.embed_init_gain : 1.0 2024-01-16 09:53:55,384 - INFO - joeynmt.config - cfg.model.tied_embeddings : True 2024-01-16 09:53:55,384 - INFO - joeynmt.config - cfg.model.tied_softmax : True 2024-01-16 09:53:55,384 - INFO - joeynmt.config - cfg.model.encoder.type : transformer 2024-01-16 09:53:55,384 - INFO - joeynmt.config - cfg.model.encoder.num_layers : 6 2024-01-16 09:53:55,384 - INFO - joeynmt.config - cfg.model.encoder.num_heads : 8 2024-01-16 09:53:55,384 - INFO - joeynmt.config - cfg.model.encoder.embeddings.embedding_dim : 1024 2024-01-16 09:53:55,384 - INFO - joeynmt.config - cfg.model.encoder.embeddings.scale : True 2024-01-16 09:53:55,384 - INFO - joeynmt.config - cfg.model.encoder.embeddings.dropout : 0.1 2024-01-16 09:53:55,384 - INFO - joeynmt.config - cfg.model.encoder.hidden_size : 1024 2024-01-16 09:53:55,384 - INFO - joeynmt.config - cfg.model.encoder.ff_size : 4096 2024-01-16 09:53:55,384 - INFO - joeynmt.config - cfg.model.encoder.dropout : 0.1 2024-01-16 09:53:55,384 - INFO - joeynmt.config - cfg.model.encoder.layer_norm : pre 2024-01-16 09:53:55,384 - INFO - joeynmt.config - cfg.model.encoder.activation : relu 2024-01-16 09:53:55,384 - INFO - joeynmt.config - cfg.model.decoder.type : transformer 2024-01-16 09:53:55,384 - INFO - joeynmt.config - cfg.model.decoder.num_layers : 6 2024-01-16 09:53:55,384 - INFO - joeynmt.config - cfg.model.decoder.num_heads : 8 2024-01-16 09:53:55,384 - INFO - joeynmt.config - cfg.model.decoder.embeddings.embedding_dim : 1024 2024-01-16 09:53:55,384 - INFO - joeynmt.config - cfg.model.decoder.embeddings.scale : True 2024-01-16 09:53:55,384 - INFO - joeynmt.config - cfg.model.decoder.embeddings.dropout : 0.1 2024-01-16 09:53:55,384 - INFO - joeynmt.config - cfg.model.decoder.hidden_size : 1024 2024-01-16 09:53:55,384 - INFO - joeynmt.config - cfg.model.decoder.ff_size : 4096 2024-01-16 09:53:55,385 - INFO - joeynmt.config - cfg.model.decoder.dropout : 0.1 2024-01-16 09:53:55,385 - INFO - joeynmt.config - cfg.model.decoder.layer_norm : pre 2024-01-16 09:53:55,385 - INFO - joeynmt.config - cfg.model.decoder.activation : relu 2024-01-16 09:53:55,385 - INFO - joeynmt.data - Building tokenizer... 2024-01-16 09:53:55,473 - INFO - joeynmt.tokenizers - src tokenizer: SentencePieceTokenizer(level=bpe, lowercase=False, normalize=False, filter_by_length=(-1, 512), pretokenizer=none, tokenizer=SentencePieceProcessor, nbest_size=5, alpha=0.0) 2024-01-16 09:53:55,473 - INFO - joeynmt.tokenizers - trg tokenizer: SentencePieceTokenizer(level=bpe, lowercase=False, normalize=False, filter_by_length=(-1, 512), pretokenizer=none, tokenizer=SentencePieceProcessor, nbest_size=5, alpha=0.0) 2024-01-16 09:53:55,473 - INFO - joeynmt.data - Loading train set... 2024-01-16 09:54:01,394 - INFO - joeynmt.data - Building vocabulary... 2024-01-16 09:54:11,386 - INFO - joeynmt.data - Loading dev set... 2024-01-16 09:54:11,433 - INFO - joeynmt.data - Loading test set... 2024-01-16 09:54:11,543 - INFO - joeynmt.data - Data loaded. 2024-01-16 09:54:11,543 - INFO - joeynmt.data - Train dataset: TsvDataset(split=train, len=702202, src_lang=src, trg_lang=trg, has_trg=True, random_subset=-1, has_src_prompt=True, has_trg_prompt=True) 2024-01-16 09:54:11,544 - INFO - joeynmt.data - Valid dataset: TsvDataset(split=dev, len=5956, src_lang=src, trg_lang=trg, has_trg=True, random_subset=500, has_src_prompt=True, has_trg_prompt=True) 2024-01-16 09:54:11,544 - INFO - joeynmt.data - Test dataset: TsvDataset(split=test, len=16896, src_lang=src, trg_lang=trg, has_trg=True, random_subset=-1, has_src_prompt=True, has_trg_prompt=True) 2024-01-16 09:54:11,545 - INFO - joeynmt.data - First training example: [SRC] ▁It ▁can ▁be ▁a ▁very ▁complicated ▁thing , ▁the ▁ocean . [TRG] ▁Ca ▁peut ▁être ▁très ▁compliqué , ▁l ' océan . 2024-01-16 09:54:11,545 - INFO - joeynmt.data - First 10 Src tokens: (0) (1) (2) (3) (4) (5) (6) (7) (8) , (9) . 2024-01-16 09:54:11,545 - INFO - joeynmt.data - First 10 Trg tokens: (0) (1) (2) (3) (4) (5) (6) (7) (8) , (9) . 2024-01-16 09:54:11,545 - INFO - joeynmt.data - Number of unique Src tokens (vocab_size): 32000 2024-01-16 09:54:11,545 - INFO - joeynmt.data - Number of unique Trg tokens (vocab_size): 32000 2024-01-16 09:54:11,561 - WARNING - joeynmt.tokenizers - models/iwslt14_prompt/sp.model already exists. Stop copying. 2024-01-16 09:54:11,565 - INFO - joeynmt.model - Building an encoder-decoder model... 2024-01-16 09:54:14,029 - INFO - joeynmt.model - Enc-dec model built. 2024-01-16 09:54:14,032 - INFO - joeynmt.model - Total params: 209129472 2024-01-16 09:54:14,033 - DEBUG - joeynmt.model - Trainable parameters: ['decoder.layer_norm.bias', 'decoder.layer_norm.weight', 'decoder.layers.0.dec_layer_norm.bias', 'decoder.layers.0.dec_layer_norm.weight', 'decoder.layers.0.feed_forward.layer_norm.bias', 'decoder.layers.0.feed_forward.layer_norm.weight', 'decoder.layers.0.feed_forward.pwff_layer.0.bias', 'decoder.layers.0.feed_forward.pwff_layer.0.weight', 'decoder.layers.0.feed_forward.pwff_layer.3.bias', 'decoder.layers.0.feed_forward.pwff_layer.3.weight', 'decoder.layers.0.src_trg_att.k_layer.bias', 'decoder.layers.0.src_trg_att.k_layer.weight', 'decoder.layers.0.src_trg_att.output_layer.bias', 'decoder.layers.0.src_trg_att.output_layer.weight', 'decoder.layers.0.src_trg_att.q_layer.bias', 'decoder.layers.0.src_trg_att.q_layer.weight', 'decoder.layers.0.src_trg_att.v_layer.bias', 'decoder.layers.0.src_trg_att.v_layer.weight', 'decoder.layers.0.trg_trg_att.k_layer.bias', 'decoder.layers.0.trg_trg_att.k_layer.weight', 'decoder.layers.0.trg_trg_att.output_layer.bias', 'decoder.layers.0.trg_trg_att.output_layer.weight', 'decoder.layers.0.trg_trg_att.q_layer.bias', 'decoder.layers.0.trg_trg_att.q_layer.weight', 'decoder.layers.0.trg_trg_att.v_layer.bias', 'decoder.layers.0.trg_trg_att.v_layer.weight', 'decoder.layers.0.x_layer_norm.bias', 'decoder.layers.0.x_layer_norm.weight', 'decoder.layers.1.dec_layer_norm.bias', 'decoder.layers.1.dec_layer_norm.weight', 'decoder.layers.1.feed_forward.layer_norm.bias', 'decoder.layers.1.feed_forward.layer_norm.weight', 'decoder.layers.1.feed_forward.pwff_layer.0.bias', 'decoder.layers.1.feed_forward.pwff_layer.0.weight', 'decoder.layers.1.feed_forward.pwff_layer.3.bias', 'decoder.layers.1.feed_forward.pwff_layer.3.weight', 'decoder.layers.1.src_trg_att.k_layer.bias', 'decoder.layers.1.src_trg_att.k_layer.weight', 'decoder.layers.1.src_trg_att.output_layer.bias', 'decoder.layers.1.src_trg_att.output_layer.weight', 'decoder.layers.1.src_trg_att.q_layer.bias', 'decoder.layers.1.src_trg_att.q_layer.weight', 'decoder.layers.1.src_trg_att.v_layer.bias', 'decoder.layers.1.src_trg_att.v_layer.weight', 'decoder.layers.1.trg_trg_att.k_layer.bias', 'decoder.layers.1.trg_trg_att.k_layer.weight', 'decoder.layers.1.trg_trg_att.output_layer.bias', 'decoder.layers.1.trg_trg_att.output_layer.weight', 'decoder.layers.1.trg_trg_att.q_layer.bias', 'decoder.layers.1.trg_trg_att.q_layer.weight', 'decoder.layers.1.trg_trg_att.v_layer.bias', 'decoder.layers.1.trg_trg_att.v_layer.weight', 'decoder.layers.1.x_layer_norm.bias', 'decoder.layers.1.x_layer_norm.weight', 'decoder.layers.2.dec_layer_norm.bias', 'decoder.layers.2.dec_layer_norm.weight', 'decoder.layers.2.feed_forward.layer_norm.bias', 'decoder.layers.2.feed_forward.layer_norm.weight', 'decoder.layers.2.feed_forward.pwff_layer.0.bias', 'decoder.layers.2.feed_forward.pwff_layer.0.weight', 'decoder.layers.2.feed_forward.pwff_layer.3.bias', 'decoder.layers.2.feed_forward.pwff_layer.3.weight', 'decoder.layers.2.src_trg_att.k_layer.bias', 'decoder.layers.2.src_trg_att.k_layer.weight', 'decoder.layers.2.src_trg_att.output_layer.bias', 'decoder.layers.2.src_trg_att.output_layer.weight', 'decoder.layers.2.src_trg_att.q_layer.bias', 'decoder.layers.2.src_trg_att.q_layer.weight', 'decoder.layers.2.src_trg_att.v_layer.bias', 'decoder.layers.2.src_trg_att.v_layer.weight', 'decoder.layers.2.trg_trg_att.k_layer.bias', 'decoder.layers.2.trg_trg_att.k_layer.weight', 'decoder.layers.2.trg_trg_att.output_layer.bias', 'decoder.layers.2.trg_trg_att.output_layer.weight', 'decoder.layers.2.trg_trg_att.q_layer.bias', 'decoder.layers.2.trg_trg_att.q_layer.weight', 'decoder.layers.2.trg_trg_att.v_layer.bias', 'decoder.layers.2.trg_trg_att.v_layer.weight', 'decoder.layers.2.x_layer_norm.bias', 'decoder.layers.2.x_layer_norm.weight', 'decoder.layers.3.dec_layer_norm.bias', 'decoder.layers.3.dec_layer_norm.weight', 'decoder.layers.3.feed_forward.layer_norm.bias', 'decoder.layers.3.feed_forward.layer_norm.weight', 'decoder.layers.3.feed_forward.pwff_layer.0.bias', 'decoder.layers.3.feed_forward.pwff_layer.0.weight', 'decoder.layers.3.feed_forward.pwff_layer.3.bias', 'decoder.layers.3.feed_forward.pwff_layer.3.weight', 'decoder.layers.3.src_trg_att.k_layer.bias', 'decoder.layers.3.src_trg_att.k_layer.weight', 'decoder.layers.3.src_trg_att.output_layer.bias', 'decoder.layers.3.src_trg_att.output_layer.weight', 'decoder.layers.3.src_trg_att.q_layer.bias', 'decoder.layers.3.src_trg_att.q_layer.weight', 'decoder.layers.3.src_trg_att.v_layer.bias', 'decoder.layers.3.src_trg_att.v_layer.weight', 'decoder.layers.3.trg_trg_att.k_layer.bias', 'decoder.layers.3.trg_trg_att.k_layer.weight', 'decoder.layers.3.trg_trg_att.output_layer.bias', 'decoder.layers.3.trg_trg_att.output_layer.weight', 'decoder.layers.3.trg_trg_att.q_layer.bias', 'decoder.layers.3.trg_trg_att.q_layer.weight', 'decoder.layers.3.trg_trg_att.v_layer.bias', 'decoder.layers.3.trg_trg_att.v_layer.weight', 'decoder.layers.3.x_layer_norm.bias', 'decoder.layers.3.x_layer_norm.weight', 'decoder.layers.4.dec_layer_norm.bias', 'decoder.layers.4.dec_layer_norm.weight', 'decoder.layers.4.feed_forward.layer_norm.bias', 'decoder.layers.4.feed_forward.layer_norm.weight', 'decoder.layers.4.feed_forward.pwff_layer.0.bias', 'decoder.layers.4.feed_forward.pwff_layer.0.weight', 'decoder.layers.4.feed_forward.pwff_layer.3.bias', 'decoder.layers.4.feed_forward.pwff_layer.3.weight', 'decoder.layers.4.src_trg_att.k_layer.bias', 'decoder.layers.4.src_trg_att.k_layer.weight', 'decoder.layers.4.src_trg_att.output_layer.bias', 'decoder.layers.4.src_trg_att.output_layer.weight', 'decoder.layers.4.src_trg_att.q_layer.bias', 'decoder.layers.4.src_trg_att.q_layer.weight', 'decoder.layers.4.src_trg_att.v_layer.bias', 'decoder.layers.4.src_trg_att.v_layer.weight', 'decoder.layers.4.trg_trg_att.k_layer.bias', 'decoder.layers.4.trg_trg_att.k_layer.weight', 'decoder.layers.4.trg_trg_att.output_layer.bias', 'decoder.layers.4.trg_trg_att.output_layer.weight', 'decoder.layers.4.trg_trg_att.q_layer.bias', 'decoder.layers.4.trg_trg_att.q_layer.weight', 'decoder.layers.4.trg_trg_att.v_layer.bias', 'decoder.layers.4.trg_trg_att.v_layer.weight', 'decoder.layers.4.x_layer_norm.bias', 'decoder.layers.4.x_layer_norm.weight', 'decoder.layers.5.dec_layer_norm.bias', 'decoder.layers.5.dec_layer_norm.weight', 'decoder.layers.5.feed_forward.layer_norm.bias', 'decoder.layers.5.feed_forward.layer_norm.weight', 'decoder.layers.5.feed_forward.pwff_layer.0.bias', 'decoder.layers.5.feed_forward.pwff_layer.0.weight', 'decoder.layers.5.feed_forward.pwff_layer.3.bias', 'decoder.layers.5.feed_forward.pwff_layer.3.weight', 'decoder.layers.5.src_trg_att.k_layer.bias', 'decoder.layers.5.src_trg_att.k_layer.weight', 'decoder.layers.5.src_trg_att.output_layer.bias', 'decoder.layers.5.src_trg_att.output_layer.weight', 'decoder.layers.5.src_trg_att.q_layer.bias', 'decoder.layers.5.src_trg_att.q_layer.weight', 'decoder.layers.5.src_trg_att.v_layer.bias', 'decoder.layers.5.src_trg_att.v_layer.weight', 'decoder.layers.5.trg_trg_att.k_layer.bias', 'decoder.layers.5.trg_trg_att.k_layer.weight', 'decoder.layers.5.trg_trg_att.output_layer.bias', 'decoder.layers.5.trg_trg_att.output_layer.weight', 'decoder.layers.5.trg_trg_att.q_layer.bias', 'decoder.layers.5.trg_trg_att.q_layer.weight', 'decoder.layers.5.trg_trg_att.v_layer.bias', 'decoder.layers.5.trg_trg_att.v_layer.weight', 'decoder.layers.5.x_layer_norm.bias', 'decoder.layers.5.x_layer_norm.weight', 'encoder.layer_norm.bias', 'encoder.layer_norm.weight', 'encoder.layers.0.feed_forward.layer_norm.bias', 'encoder.layers.0.feed_forward.layer_norm.weight', 'encoder.layers.0.feed_forward.pwff_layer.0.bias', 'encoder.layers.0.feed_forward.pwff_layer.0.weight', 'encoder.layers.0.feed_forward.pwff_layer.3.bias', 'encoder.layers.0.feed_forward.pwff_layer.3.weight', 'encoder.layers.0.layer_norm.bias', 'encoder.layers.0.layer_norm.weight', 'encoder.layers.0.src_src_att.k_layer.bias', 'encoder.layers.0.src_src_att.k_layer.weight', 'encoder.layers.0.src_src_att.output_layer.bias', 'encoder.layers.0.src_src_att.output_layer.weight', 'encoder.layers.0.src_src_att.q_layer.bias', 'encoder.layers.0.src_src_att.q_layer.weight', 'encoder.layers.0.src_src_att.v_layer.bias', 'encoder.layers.0.src_src_att.v_layer.weight', 'encoder.layers.1.feed_forward.layer_norm.bias', 'encoder.layers.1.feed_forward.layer_norm.weight', 'encoder.layers.1.feed_forward.pwff_layer.0.bias', 'encoder.layers.1.feed_forward.pwff_layer.0.weight', 'encoder.layers.1.feed_forward.pwff_layer.3.bias', 'encoder.layers.1.feed_forward.pwff_layer.3.weight', 'encoder.layers.1.layer_norm.bias', 'encoder.layers.1.layer_norm.weight', 'encoder.layers.1.src_src_att.k_layer.bias', 'encoder.layers.1.src_src_att.k_layer.weight', 'encoder.layers.1.src_src_att.output_layer.bias', 'encoder.layers.1.src_src_att.output_layer.weight', 'encoder.layers.1.src_src_att.q_layer.bias', 'encoder.layers.1.src_src_att.q_layer.weight', 'encoder.layers.1.src_src_att.v_layer.bias', 'encoder.layers.1.src_src_att.v_layer.weight', 'encoder.layers.2.feed_forward.layer_norm.bias', 'encoder.layers.2.feed_forward.layer_norm.weight', 'encoder.layers.2.feed_forward.pwff_layer.0.bias', 'encoder.layers.2.feed_forward.pwff_layer.0.weight', 'encoder.layers.2.feed_forward.pwff_layer.3.bias', 'encoder.layers.2.feed_forward.pwff_layer.3.weight', 'encoder.layers.2.layer_norm.bias', 'encoder.layers.2.layer_norm.weight', 'encoder.layers.2.src_src_att.k_layer.bias', 'encoder.layers.2.src_src_att.k_layer.weight', 'encoder.layers.2.src_src_att.output_layer.bias', 'encoder.layers.2.src_src_att.output_layer.weight', 'encoder.layers.2.src_src_att.q_layer.bias', 'encoder.layers.2.src_src_att.q_layer.weight', 'encoder.layers.2.src_src_att.v_layer.bias', 'encoder.layers.2.src_src_att.v_layer.weight', 'encoder.layers.3.feed_forward.layer_norm.bias', 'encoder.layers.3.feed_forward.layer_norm.weight', 'encoder.layers.3.feed_forward.pwff_layer.0.bias', 'encoder.layers.3.feed_forward.pwff_layer.0.weight', 'encoder.layers.3.feed_forward.pwff_layer.3.bias', 'encoder.layers.3.feed_forward.pwff_layer.3.weight', 'encoder.layers.3.layer_norm.bias', 'encoder.layers.3.layer_norm.weight', 'encoder.layers.3.src_src_att.k_layer.bias', 'encoder.layers.3.src_src_att.k_layer.weight', 'encoder.layers.3.src_src_att.output_layer.bias', 'encoder.layers.3.src_src_att.output_layer.weight', 'encoder.layers.3.src_src_att.q_layer.bias', 'encoder.layers.3.src_src_att.q_layer.weight', 'encoder.layers.3.src_src_att.v_layer.bias', 'encoder.layers.3.src_src_att.v_layer.weight', 'encoder.layers.4.feed_forward.layer_norm.bias', 'encoder.layers.4.feed_forward.layer_norm.weight', 'encoder.layers.4.feed_forward.pwff_layer.0.bias', 'encoder.layers.4.feed_forward.pwff_layer.0.weight', 'encoder.layers.4.feed_forward.pwff_layer.3.bias', 'encoder.layers.4.feed_forward.pwff_layer.3.weight', 'encoder.layers.4.layer_norm.bias', 'encoder.layers.4.layer_norm.weight', 'encoder.layers.4.src_src_att.k_layer.bias', 'encoder.layers.4.src_src_att.k_layer.weight', 'encoder.layers.4.src_src_att.output_layer.bias', 'encoder.layers.4.src_src_att.output_layer.weight', 'encoder.layers.4.src_src_att.q_layer.bias', 'encoder.layers.4.src_src_att.q_layer.weight', 'encoder.layers.4.src_src_att.v_layer.bias', 'encoder.layers.4.src_src_att.v_layer.weight', 'encoder.layers.5.feed_forward.layer_norm.bias', 'encoder.layers.5.feed_forward.layer_norm.weight', 'encoder.layers.5.feed_forward.pwff_layer.0.bias', 'encoder.layers.5.feed_forward.pwff_layer.0.weight', 'encoder.layers.5.feed_forward.pwff_layer.3.bias', 'encoder.layers.5.feed_forward.pwff_layer.3.weight', 'encoder.layers.5.layer_norm.bias', 'encoder.layers.5.layer_norm.weight', 'encoder.layers.5.src_src_att.k_layer.bias', 'encoder.layers.5.src_src_att.k_layer.weight', 'encoder.layers.5.src_src_att.output_layer.bias', 'encoder.layers.5.src_src_att.output_layer.weight', 'encoder.layers.5.src_src_att.q_layer.bias', 'encoder.layers.5.src_src_att.q_layer.weight', 'encoder.layers.5.src_src_att.v_layer.bias', 'encoder.layers.5.src_src_att.v_layer.weight', 'src_embed.lut.weight'] 2024-01-16 09:54:15,518 - INFO - joeynmt.prediction - DataParallelWrapper( (module): DistributedDataParallel( (module): Model( encoder=TransformerEncoder(num_layers=6, num_heads=8, alpha=1.0, layer_norm="pre", activation=ReLU()), decoder=TransformerDecoder(num_layers=6, num_heads=8, alpha=1.0, layer_norm="pre", activation=ReLU()), src_embed=Embeddings(embedding_dim=1024, vocab_size=32000), trg_embed=Embeddings(embedding_dim=1024, vocab_size=32000), loss_function=XentLoss(criterion=KLDivLoss(), smoothing=0.1)) ) ) 2024-01-16 09:54:29,093 - INFO - joeynmt.builders - AdamW(lr=0.0002, weight_decay=0.001, betas=[0.9, 0.98]) 2024-01-16 09:54:29,094 - INFO - joeynmt.builders - WarmupInverseSquareRootScheduler(warmup=10000, decay_rate=0.020000, peak_rate=0.0002, min_rate=1e-07) 2024-01-16 09:54:29,094 - INFO - joeynmt.training - Train config: device: cuda n_gpu: 2 ddp training: True 16-bits training: True gradient accumulation: 4 batch size per device: 32 effective batch size (w. parallel & accumulation): 256 2024-01-16 09:54:29,095 - INFO - joeynmt.training - EPOCH 1 2024-01-16 09:55:46,611 - INFO - joeynmt.training - Epoch 1, Step: 100, Batch Loss: 8.358168, Batch Acc: 0.050132, Tokens per Sec: 16196, Lr: 0.000002 2024-01-16 09:57:03,484 - INFO - joeynmt.training - Epoch 1, Step: 200, Batch Loss: 8.017015, Batch Acc: 0.053869, Tokens per Sec: 16246, Lr: 0.000004 2024-01-16 09:58:19,762 - INFO - joeynmt.training - Epoch 1, Step: 300, Batch Loss: 7.480748, Batch Acc: 0.068775, Tokens per Sec: 16332, Lr: 0.000006 2024-01-16 09:59:36,625 - INFO - joeynmt.training - Epoch 1, Step: 400, Batch Loss: 6.797713, Batch Acc: 0.099258, Tokens per Sec: 16244, Lr: 0.000008 2024-01-16 10:00:54,580 - INFO - joeynmt.training - Epoch 1, Step: 500, Batch Loss: 6.280566, Batch Acc: 0.118143, Tokens per Sec: 16142, Lr: 0.000010 2024-01-16 10:02:10,271 - INFO - joeynmt.training - Epoch 1, Step: 600, Batch Loss: 5.890950, Batch Acc: 0.136770, Tokens per Sec: 16507, Lr: 0.000012 2024-01-16 10:03:27,401 - INFO - joeynmt.training - Epoch 1, Step: 700, Batch Loss: 5.555294, Batch Acc: 0.157129, Tokens per Sec: 16190, Lr: 0.000014 2024-01-16 10:04:44,320 - INFO - joeynmt.training - Epoch 1, Step: 800, Batch Loss: 5.375746, Batch Acc: 0.170306, Tokens per Sec: 16249, Lr: 0.000016 2024-01-16 10:06:01,249 - INFO - joeynmt.training - Epoch 1, Step: 900, Batch Loss: 5.187682, Batch Acc: 0.179104, Tokens per Sec: 16233, Lr: 0.000018 2024-01-16 10:07:16,906 - INFO - joeynmt.training - Epoch 1, Step: 1000, Batch Loss: 5.025946, Batch Acc: 0.183847, Tokens per Sec: 16478, Lr: 0.000020 2024-01-16 10:07:16,907 - INFO - joeynmt.datasets - Sample random subset from dev data: n=500, seed=1042 2024-01-16 10:07:16,907 - INFO - joeynmt.prediction - Predicting 500 example(s)... (Greedy decoding with min_output_length=1, max_output_length=512, return_prob='none', generate_unk=True, repetition_penalty=-1, no_repeat_ngram_size=-1) 2024-01-16 10:08:13,644 - INFO - joeynmt.prediction - Generation took 56.7287[sec]. 2024-01-16 10:08:13,809 - INFO - joeynmt.metrics - nrefs:1|case:lc|eff:no|tok:13a|smooth:exp|version:2.4.0 2024-01-16 10:08:13,810 - INFO - joeynmt.prediction - Evaluation result (greedy): bleu: 0.02, loss: 5.33, ppl: 207.10, acc: 0.14, 0.1244[sec] 2024-01-16 10:08:13,810 - INFO - joeynmt.training - Hooray! New best validation result [bleu]! 2024-01-16 10:08:16,690 - INFO - joeynmt.training - Checkpoint saved in models/iwslt14_prompt/1000.ckpt. 2024-01-16 10:08:16,694 - INFO - joeynmt.training - Example #0 2024-01-16 10:08:16,695 - INFO - joeynmt.training - Source: All of these peoples teach us that there are other ways of being, other ways of thinking, other ways of orienting yourself in the Earth. 2024-01-16 10:08:16,695 - INFO - joeynmt.training - Reference: Tous ces peuples nous enseignent qu'il y a d'autres façons d'être, d'autres façons de penser, d'autres manières de s'orienter sur Terre. 2024-01-16 10:08:16,695 - INFO - joeynmt.training - Hypothesis: Et la la la la la la la la la la la la la la la la la la la la la la la la la la la la la la la la la la la la la la la la la la la la la la la la la la la la la la la la la la la la la la la la la la la la la la la la la la la la la la la la la la la la la la la la la la la la la la la la la la la la la la la la la la la la la la la la la la la la la la la la la la la la la la la la la la la la 2024-01-16 10:08:16,696 - DEBUG - joeynmt.training - Tokenized source: ['▁All', '▁of', '▁these', '▁people', 's', '▁teach', '▁us', '▁that', '▁there', '▁are', '▁other', '▁ways', '▁of', '▁being', ',', '▁other', '▁ways', '▁of', '▁thinking', ',', '▁other', '▁ways', '▁of', '▁orient', 'ing', '▁yourself', '▁in', '▁the', '▁Earth', '.'] 2024-01-16 10:08:16,696 - DEBUG - joeynmt.training - Tokenized reference: ['▁Tous', '▁ces', '▁peuple', 's', '▁nous', '▁enseigne', 'nt', '▁qu', "'", 'il', '▁y', '▁a', '▁d', "'", 'autres', '▁façons', '▁d', "'", 'être', ',', '▁d', "'", 'autres', '▁façons', '▁de', '▁penser', ',', '▁d', "'", 'autres', '▁manières', '▁de', '▁s', "'", 'orient', 'er', '▁sur', '▁Terre', '.'] 2024-01-16 10:08:16,696 - DEBUG - joeynmt.training - Tokenized hypothesis: ['', '▁Et', '▁que', '▁ce', '▁soit', '▁le', '▁Pen', 'an', '▁dans', '▁les', '▁forêts', '▁du', '▁Bo', 'r', 'né', 'o', ',', '▁ou', '▁les', '▁a', 'co', 'ly', 'tes', '▁V', 'oodoo', '▁à', '▁Haïti', ',', '▁ou', '▁bien', '▁les', '▁guerrier', 's', '▁dans', '▁le', '▁désert', '▁du', '▁K', 'ais', 'ut', '▁au', '▁nord', '▁du', '▁Kenya', ',', '▁le', '▁C', 'ur', 'ende', 'ro', '▁dans', '▁les', '▁montagnes', '▁des', '▁And', 'es', ',', '▁ou', '▁bien', '▁un', '▁car', 'ava', 'n', 's', 'é', 'ra', 'il', '▁en', '▁plein', '▁milieu', '▁du', '▁Sahara', '.', '▁A', '▁propos', ',', '▁c', "'", 'est', '▁la', '▁personne', '▁avec', '▁qui', '▁j', "'", 'ai', '▁voyagé', '▁dans', '▁le', '▁désert', '▁il', '▁y', '▁un', '▁mois', ',', '▁ou', '▁effectivement', ',', '▁le', '▁gardien', '▁de', '▁troupeau', '▁de', '▁Yak', 's', '▁sur', '▁les', '▁fla', 'n', 'c', 's', '▁du', '▁Q', 'o', 'mo', 'lang', 'ma', ',', '▁l', "'", 'Everest', ',', '▁la', '▁dé', 'esse', '▁du', '▁monde', '.', '', '▁Et', '▁la', '▁la', '▁la', '▁la', '▁la', '▁la', '▁la', '▁la', '▁la', '▁la', '▁la', '▁la', '▁la', '▁la', '▁la', '▁la', '▁la', '▁la', '▁la', '▁la', '▁la', '▁la', '▁la', '▁la', '▁la', '▁la', '▁la', '▁la', '▁la', '▁la', '▁la', '▁la', '▁la', '▁la', '▁la', '▁la', '▁la', '▁la', '▁la', '▁la', '▁la', '▁la', '▁la', '▁la', '▁la', '▁la', '▁la', '▁la', '▁la', '▁la', '▁la', '▁la', '▁la', '▁la', '▁la', '▁la', '▁la', '▁la', '▁la', '▁la', '▁la', '▁la', '▁la', '▁la', '▁la', '▁la', '▁la', '▁la', '▁la', '▁la', '▁la', '▁la', '▁la', '▁la', '▁la', '▁la', '▁la', '▁la', '▁la', '▁la', '▁la', '▁la', '▁la', '▁la', '▁la', '▁la', '▁la', '▁la', '▁la', '▁la', '▁la', '▁la', '▁la', '▁la', '▁la', '▁la', '▁la', '▁la', '▁la', '▁la', '▁la', '▁la', '▁la', '▁la', '▁la', '▁la', '▁la', '▁la', '▁la', '▁la', '▁la', '▁la', '▁la', '▁la', '▁la', '▁la', '▁la', '▁la', '▁la', '▁la', '▁la', '▁la', '▁la', '▁la', '▁la', '▁la', '▁la', '▁la', '▁la', '▁la', '▁la', '▁la'] 2024-01-16 10:08:16,696 - INFO - joeynmt.training - Example #1 2024-01-16 10:08:16,696 - INFO - joeynmt.training - Source: Et ceci est une idée, si on y réfléchit, qui ne peut que vous remplir d'espoir. 2024-01-16 10:08:16,696 - INFO - joeynmt.training - Reference: And this is an idea, if you think about it, can only fill you with hope. 2024-01-16 10:08:16,697 - INFO - joeynmt.training - Hypothesis: And I's, and I's a the the the the the the the the the the the the the the the the the the the the the the the the the the the the the the the the the the the the the the the the the the the the the the the the the the the the the. 2024-01-16 10:08:16,697 - DEBUG - joeynmt.training - Tokenized source: ['▁Et', '▁ceci', '▁est', '▁une', '▁idée', ',', '▁si', '▁on', '▁y', '▁réfléchi', 't', ',', '▁qui', '▁ne', '▁peut', '▁que', '▁vous', '▁remplir', '▁d', "'", 'espoir', '.'] 2024-01-16 10:08:16,697 - DEBUG - joeynmt.training - Tokenized reference: ['▁And', '▁this', '▁is', '▁an', '▁idea', ',', '▁if', '▁you', '▁think', '▁about', '▁it', ',', '▁can', '▁only', '▁fill', '▁you', '▁with', '▁hope', '.'] 2024-01-16 10:08:16,697 - DEBUG - joeynmt.training - Tokenized hypothesis: ['', '▁All', '▁of', '▁these', '▁people', 's', '▁teach', '▁us', '▁that', '▁there', '▁are', '▁other', '▁ways', '▁of', '▁being', ',', '▁other', '▁ways', '▁of', '▁thinking', ',', '▁other', '▁ways', '▁of', '▁orient', 'ing', '▁yourself', '▁in', '▁the', '▁Earth', '.', '', '▁And', '▁I', "'", 's', ',', '▁and', '▁I', "'", 's', '▁a', '▁the', '▁the', '▁the', '▁the', '▁the', '▁the', '▁the', '▁the', '▁the', '▁the', '▁the', '▁the', '▁the', '▁the', '▁the', '▁the', '▁the', '▁the', '▁the', '▁the', '▁the', '▁the', '▁the', '▁the', '▁the', '▁the', '▁the', '▁the', '▁the', '▁the', '▁the', '▁the', '▁the', '▁the', '▁the', '▁the', '▁the', '▁the', '▁the', '▁the', '▁the', '▁the', '▁the', '▁the', '▁the', '▁the', '▁the', '▁the', '▁the', '▁the', '▁the', '▁the', '▁the', '.'] 2024-01-16 10:08:16,697 - INFO - joeynmt.training - Example #2 2024-01-16 10:08:16,698 - INFO - joeynmt.training - Source: It's the symbol of all that we are and all that we can be as an astonishingly inquisitive species. 2024-01-16 10:08:16,698 - INFO - joeynmt.training - Reference: C'est le symbole de tout ce que nous sommes et tout ce que nous pouvons être en tant qu'espèce dotée d'une curiosité stupéfiante. 2024-01-16 10:08:16,698 - INFO - joeynmt.training - Hypothesis: Et la la la la la la la la la la la la la la la la la la la la la la la la la la la la la la la la la la la la la la la la la la la la la la la la la la la la la la la la la la la la la la la la la la la la la la la la la la la la la la la la la la la la la 2024-01-16 10:08:16,699 - DEBUG - joeynmt.training - Tokenized source: ['▁It', "'", 's', '▁the', '▁symbol', '▁of', '▁all', '▁that', '▁we', '▁are', '▁and', '▁all', '▁that', '▁we', '▁can', '▁be', '▁as', '▁an', '▁astonishing', 'ly', '▁in', 'qui', 's', 'it', 'ive', '▁species', '.'] 2024-01-16 10:08:16,699 - DEBUG - joeynmt.training - Tokenized reference: ['▁C', "'", 'est', '▁le', '▁symbole', '▁de', '▁tout', '▁ce', '▁que', '▁nous', '▁sommes', '▁et', '▁tout', '▁ce', '▁que', '▁nous', '▁pouvons', '▁être', '▁en', '▁tant', '▁qu', "'", 'espèce', '▁do', 't', 'ée', '▁d', "'", 'une', '▁curiosité', '▁stupéfiant', 'e', '.'] 2024-01-16 10:08:16,699 - DEBUG - joeynmt.training - Tokenized hypothesis: ['', '▁L', "'", 'e', 'th', 'no', 's', 'ph', 'ère', '▁est', '▁l', "'", 'héritage', '▁de', '▁l', "'", 'humanité', '.', '', '▁Et', '▁la', '▁la', '▁la', '▁la', '▁la', '▁la', '▁la', '▁la', '▁la', '▁la', '▁la', '▁la', '▁la', '▁la', '▁la', '▁la', '▁la', '▁la', '▁la', '▁la', '▁la', '▁la', '▁la', '▁la', '▁la', '▁la', '▁la', '▁la', '▁la', '▁la', '▁la', '▁la', '▁la', '▁la', '▁la', '▁la', '▁la', '▁la', '▁la', '▁la', '▁la', '▁la', '▁la', '▁la', '▁la', '▁la', '▁la', '▁la', '▁la', '▁la', '▁la', '▁la', '▁la', '▁la', '▁la', '▁la', '▁la', '▁la', '▁la', '▁la', '▁la', '▁la', '▁la', '▁la', '▁la', '▁la', '▁la', '▁la', '▁la', '▁la', '▁la', '▁la', '▁la', '▁la', '▁la', '▁la', '▁la', '▁la', '▁la', '▁la', '▁la', '▁la', '▁la', '▁la', '▁la'] 2024-01-16 10:08:16,699 - INFO - joeynmt.training - Example #3 2024-01-16 10:08:16,699 - INFO - joeynmt.training - Source: Une langue est une étincelle de l'esprit humain. 2024-01-16 10:08:16,699 - INFO - joeynmt.training - Reference: A language is a flash of the human spirit. 2024-01-16 10:08:16,700 - INFO - joeynmt.training - Hypothesis: And I's a the the the the the the the the the the the the the the the the the the the the the the the the the the the the the the the the the the the the. 2024-01-16 10:08:16,700 - DEBUG - joeynmt.training - Tokenized source: ['▁Une', '▁langue', '▁est', '▁une', '▁', 'étincelle', '▁de', '▁l', "'", 'esprit', '▁humain', '.'] 2024-01-16 10:08:16,700 - DEBUG - joeynmt.training - Tokenized reference: ['▁A', '▁language', '▁is', '▁a', '▁flash', '▁of', '▁the', '▁human', '▁spirit', '.'] 2024-01-16 10:08:16,700 - DEBUG - joeynmt.training - Tokenized hypothesis: ['', '▁Now', ',', '▁a', '▁language', '▁is', '▁not', '▁just', '▁a', '▁body', '▁of', '▁vocabulary', '▁or', '▁a', '▁set', '▁of', '▁grammatical', '▁rules', '.', '', '▁And', '▁I', "'", 's', '▁a', '▁the', '▁the', '▁the', '▁the', '▁the', '▁the', '▁the', '▁the', '▁the', '▁the', '▁the', '▁the', '▁the', '▁the', '▁the', '▁the', '▁the', '▁the', '▁the', '▁the', '▁the', '▁the', '▁the', '▁the', '▁the', '▁the', '▁the', '▁the', '▁the', '▁the', '▁the', '▁the', '▁the', '▁the', '▁the', '▁the', '.'] 2024-01-16 10:09:32,305 - INFO - joeynmt.training - Epoch 1, Step: 1100, Batch Loss: 5.012686, Batch Acc: 0.187792, Tokens per Sec: 16491, Lr: 0.000022 2024-01-16 10:10:48,280 - INFO - joeynmt.training - Epoch 1, Step: 1200, Batch Loss: 4.795453, Batch Acc: 0.193757, Tokens per Sec: 16465, Lr: 0.000024 2024-01-16 10:12:04,260 - INFO - joeynmt.training - Epoch 1, Step: 1300, Batch Loss: 4.788570, Batch Acc: 0.200276, Tokens per Sec: 16533, Lr: 0.000026 2024-01-16 10:13:20,503 - INFO - joeynmt.training - Epoch 1, Step: 1400, Batch Loss: 4.686733, Batch Acc: 0.206386, Tokens per Sec: 16416, Lr: 0.000028 2024-01-16 10:14:36,138 - INFO - joeynmt.training - Epoch 1, Step: 1500, Batch Loss: 4.719732, Batch Acc: 0.212446, Tokens per Sec: 16491, Lr: 0.000030 2024-01-16 10:15:51,811 - INFO - joeynmt.training - Epoch 1, Step: 1600, Batch Loss: 4.550242, Batch Acc: 0.218058, Tokens per Sec: 16510, Lr: 0.000032 2024-01-16 10:17:07,684 - INFO - joeynmt.training - Epoch 1, Step: 1700, Batch Loss: 4.546413, Batch Acc: 0.223189, Tokens per Sec: 16466, Lr: 0.000034 2024-01-16 10:18:22,820 - INFO - joeynmt.training - Epoch 1, Step: 1800, Batch Loss: 4.473071, Batch Acc: 0.230033, Tokens per Sec: 16553, Lr: 0.000036 2024-01-16 10:19:38,912 - INFO - joeynmt.training - Epoch 1, Step: 1900, Batch Loss: 4.395797, Batch Acc: 0.235364, Tokens per Sec: 16376, Lr: 0.000038 2024-01-16 10:20:54,358 - INFO - joeynmt.training - Epoch 1, Step: 2000, Batch Loss: 4.338891, Batch Acc: 0.241419, Tokens per Sec: 16604, Lr: 0.000040 2024-01-16 10:20:54,359 - INFO - joeynmt.datasets - Sample random subset from dev data: n=500, seed=2042 2024-01-16 10:20:54,359 - INFO - joeynmt.prediction - Predicting 500 example(s)... (Greedy decoding with min_output_length=1, max_output_length=512, return_prob='none', generate_unk=True, repetition_penalty=-1, no_repeat_ngram_size=-1) 2024-01-16 10:21:17,230 - INFO - joeynmt.prediction - Generation took 22.8630[sec]. 2024-01-16 10:21:17,325 - INFO - joeynmt.metrics - nrefs:1|case:lc|eff:no|tok:13a|smooth:exp|version:2.4.0 2024-01-16 10:21:17,326 - INFO - joeynmt.prediction - Evaluation result (greedy): bleu: 0.56, loss: 4.71, ppl: 111.14, acc: 0.19, 0.0734[sec] 2024-01-16 10:21:17,327 - INFO - joeynmt.training - Hooray! New best validation result [bleu]! 2024-01-16 10:21:20,265 - INFO - joeynmt.training - Checkpoint saved in models/iwslt14_prompt/2000.ckpt. 2024-01-16 10:21:20,268 - INFO - joeynmt.training - Example #0 2024-01-16 10:21:20,269 - INFO - joeynmt.training - Source: Le fait de savoir que les Jaguar shaman voyagent toujours au-delà de la voie lactée, ou que les mythes des anciens Inuit résonnent encore de sens, ou bien que dans l'Himalaya, les Bouddhistes continuent à rechercher le souffle du Dharma, c'est se rappeler de la révélation essentielle de l'anthropologie, et cela veut dire que le monde dans lequel nous vivons n'existe pas dans un sens absolu, mais est uniquement un exemple de réalité, la conséquence d'un ensemble spécifique de choix adaptés établis par notre lignée avec succès, il y a plusieurs générations. 2024-01-16 10:21:20,269 - INFO - joeynmt.training - Reference: Just to know that Jaguar shamans still journey beyond the Milky Way, or the myths of the Inuit elders still resonate with meaning, or that in the Himalaya, the Buddhists still pursue the breath of the Dharma, is to really remember the central revelation of anthropology, and that is the idea that the world in which we live does not exist in some absolute sense, but is just one model of reality, the consequence of one particular set of adaptive choices that our lineage made, albeit successfully, many generations ago. 2024-01-16 10:21:20,269 - INFO - joeynmt.training - Hypothesis: And the world's a lot of the world, and the world, the world, the world, the world's going to the world, and the world, the world, the world, the world's going to the world, the world, and the world's the world, the world, the world's the world, the world, and the world, the world, the world, the world's the world, the world, the world, the world, the world is the world, the world, the world, the world, the world, the world's the world's the world, the world, the world, the world, the world, the world's to the world's to the world, the world is the world, the world is the world, the world, the world, the world, the world, and the world, the world is the world is the world, the world. 2024-01-16 10:21:20,270 - DEBUG - joeynmt.training - Tokenized source: ['▁Le', '▁fait', '▁de', '▁savoir', '▁que', '▁les', '▁Ja', 'gu', 'ar', '▁shaman', '▁voyage', 'nt', '▁toujours', '▁au', '-', 'delà', '▁de', '▁la', '▁voie', '▁l', 'act', 'ée', ',', '▁ou', '▁que', '▁les', '▁mythe', 's', '▁des', '▁anciens', '▁Inuit', '▁résonne', 'nt', '▁encore', '▁de', '▁sens', ',', '▁ou', '▁bien', '▁que', '▁dans', '▁l', "'", 'Himalaya', ',', '▁les', '▁Bo', 'ud', 'd', 'h', 'istes', '▁continuent', '▁à', '▁recherche', 'r', '▁le', '▁souffle', '▁du', '▁D', 'ha', 'r', 'ma', ',', '▁c', "'", 'est', '▁se', '▁rappeler', '▁de', '▁la', '▁révélation', '▁essentielle', '▁de', '▁l', "'", 'anthrop', 'ologie', ',', '▁et', '▁cela', '▁veut', '▁dire', '▁que', '▁le', '▁monde', '▁dans', '▁lequel', '▁nous', '▁vivons', '▁n', "'", 'existe', '▁pas', '▁dans', '▁un', '▁sens', '▁absolu', ',', '▁mais', '▁est', '▁uniquement', '▁un', '▁exemple', '▁de', '▁réalité', ',', '▁la', '▁conséquence', '▁d', "'", 'un', '▁ensemble', '▁spécifique', '▁de', '▁choix', '▁adapté', 's', '▁établi', 's', '▁par', '▁notre', '▁lignée', '▁avec', '▁succès', ',', '▁il', '▁y', '▁a', '▁plusieurs', '▁générations', '.'] 2024-01-16 10:21:20,270 - DEBUG - joeynmt.training - Tokenized reference: ['▁Just', '▁to', '▁know', '▁that', '▁Ja', 'gu', 'ar', '▁shaman', 's', '▁still', '▁journey', '▁beyond', '▁the', '▁Milk', 'y', '▁Way', ',', '▁or', '▁the', '▁myth', 's', '▁of', '▁the', '▁Inuit', '▁el', 'der', 's', '▁still', '▁resonate', '▁with', '▁meaning', ',', '▁or', '▁that', '▁in', '▁the', '▁Himalaya', ',', '▁the', '▁Buddhist', 's', '▁still', '▁pursue', '▁the', '▁breath', '▁of', '▁the', '▁D', 'ha', 'r', 'ma', ',', '▁is', '▁to', '▁really', '▁remember', '▁the', '▁central', '▁revelation', '▁of', '▁anthropology', ',', '▁and', '▁that', '▁is', '▁the', '▁idea', '▁that', '▁the', '▁world', '▁in', '▁which', '▁we', '▁live', '▁does', '▁not', '▁exist', '▁in', '▁some', '▁absolute', '▁sense', ',', '▁but', '▁is', '▁just', '▁one', '▁model', '▁of', '▁reality', ',', '▁the', '▁consequence', '▁of', '▁one', '▁particular', '▁set', '▁of', '▁adaptive', '▁choices', '▁that', '▁our', '▁line', 'age', '▁made', ',', '▁albeit', '▁successfully', ',', '▁many', '▁generations', '▁ago', '.'] 2024-01-16 10:21:20,270 - DEBUG - joeynmt.training - Tokenized hypothesis: ['', '▁You', '▁know', ',', '▁one', '▁of', '▁the', '▁intense', '▁pleasure', 's', '▁of', '▁travel', '▁and', '▁one', '▁of', '▁the', '▁de', 'light', 's', '▁of', '▁et', 'h', 'n', 'ographic', '▁research', '▁is', '▁the', '▁opportunity', '▁to', '▁live', '▁among', 'st', '▁those', '▁who', '▁have', '▁not', '▁forgotten', '▁the', '▁old', '▁ways', ',', '▁who', '▁still', '▁feel', '▁their', '▁past', '▁in', '▁the', '▁wind', ',', '▁touch', '▁it', '▁in', '▁stones', '▁polish', 'ed', '▁by', '▁rain', ',', '▁taste', '▁it', '▁in', '▁the', '▁bitter', '▁leaves', '▁of', '▁plants', '.', '', '▁And', '▁the', '▁world', "'", 's', '▁a', '▁lot', '▁of', '▁the', '▁world', ',', '▁and', '▁the', '▁world', ',', '▁the', '▁world', ',', '▁the', '▁world', ',', '▁the', '▁world', "'", 's', '▁going', '▁to', '▁the', '▁world', ',', '▁and', '▁the', '▁world', ',', '▁the', '▁world', ',', '▁the', '▁world', ',', '▁the', '▁world', "'", 's', '▁going', '▁to', '▁the', '▁world', ',', '▁the', '▁world', ',', '▁and', '▁the', '▁world', "'", 's', '▁the', '▁world', ',', '▁the', '▁world', ',', '▁the', '▁world', "'", 's', '▁the', '▁world', ',', '▁the', '▁world', ',', '▁and', '▁the', '▁world', ',', '▁the', '▁world', ',', '▁the', '▁world', ',', '▁the', '▁world', "'", 's', '▁the', '▁world', ',', '▁the', '▁world', ',', '▁the', '▁world', ',', '▁the', '▁world', ',', '▁the', '▁world', '▁is', '▁the', '▁world', ',', '▁the', '▁world', ',', '▁the', '▁world', ',', '▁the', '▁world', ',', '▁the', '▁world', ',', '▁the', '▁world', "'", 's', '▁the', '▁world', "'", 's', '▁the', '▁world', ',', '▁the', '▁world', ',', '▁the', '▁world', ',', '▁the', '▁world', ',', '▁the', '▁world', ',', '▁the', '▁world', "'", 's', '▁to', '▁the', '▁world', "'", 's', '▁to', '▁the', '▁world', ',', '▁the', '▁world', '▁is', '▁the', '▁world', ',', '▁the', '▁world', '▁is', '▁the', '▁world', ',', '▁the', '▁world', ',', '▁the', '▁world', ',', '▁the', '▁world', ',', '▁the', '▁world', ',', '▁and', '▁the', '▁world', ',', '▁the', '▁world', '▁is', '▁the', '▁world', '▁is', '▁the', '▁world', ',', '▁the', '▁world', '.'] 2024-01-16 10:21:20,271 - INFO - joeynmt.training - Example #1 2024-01-16 10:21:20,271 - INFO - joeynmt.training - Source: Nous devons faire face à la séparation inexorable de la mort, cela ne devrait donc pas nous surprendre que nous chantions, nous dansions, nous sommes tous des artistes. 2024-01-16 10:21:20,271 - INFO - joeynmt.training - Reference: We have to deal with the inexorable separation of death, so it shouldn't surprise us that we all sing, we all dance, we all have art. 2024-01-16 10:21:20,271 - INFO - joeynmt.training - Hypothesis: And we're going to be able to be able to be able to the world, we're going to do that we're going to do that we're going to do that we're going to do. 2024-01-16 10:21:20,272 - DEBUG - joeynmt.training - Tokenized source: ['▁Nous', '▁devons', '▁faire', '▁face', '▁à', '▁la', '▁séparation', '▁inexorable', '▁de', '▁la', '▁mort', ',', '▁cela', '▁ne', '▁devrait', '▁donc', '▁pas', '▁nous', '▁surprend', 're', '▁que', '▁nous', '▁chant', 'ions', ',', '▁nous', '▁dans', 'ions', ',', '▁nous', '▁sommes', '▁tous', '▁des', '▁artistes', '.'] 2024-01-16 10:21:20,272 - DEBUG - joeynmt.training - Tokenized reference: ['▁We', '▁have', '▁to', '▁deal', '▁with', '▁the', '▁inexorable', '▁separation', '▁of', '▁death', ',', '▁so', '▁it', '▁should', 'n', "'", 't', '▁surprise', '▁us', '▁that', '▁we', '▁all', '▁sing', ',', '▁we', '▁all', '▁dance', ',', '▁we', '▁all', '▁have', '▁art', '.'] 2024-01-16 10:21:20,272 - DEBUG - joeynmt.training - Tokenized hypothesis: ['', '▁We', '▁go', '▁through', '▁initiat', 'ion', '▁r', 'ites', '.', '', '▁And', '▁we', "'", 're', '▁going', '▁to', '▁be', '▁able', '▁to', '▁be', '▁able', '▁to', '▁be', '▁able', '▁to', '▁the', '▁world', ',', '▁we', "'", 're', '▁going', '▁to', '▁do', '▁that', '▁we', "'", 're', '▁going', '▁to', '▁do', '▁that', '▁we', "'", 're', '▁going', '▁to', '▁do', '▁that', '▁we', "'", 're', '▁going', '▁to', '▁do', '.'] 2024-01-16 10:21:20,272 - INFO - joeynmt.training - Example #2 2024-01-16 10:21:20,272 - INFO - joeynmt.training - Source: All of these peoples teach us that there are other ways of being, other ways of thinking, other ways of orienting yourself in the Earth. 2024-01-16 10:21:20,272 - INFO - joeynmt.training - Reference: Tous ces peuples nous enseignent qu'il y a d'autres façons d'être, d'autres façons de penser, d'autres manières de s'orienter sur Terre. 2024-01-16 10:21:20,273 - INFO - joeynmt.training - Hypothesis: Et il y a été un monde, et la vie, et la vie, la vie, la vie, la vie est un monde, et la vie de la vie de la vie de la vie de la vie de la vie de la vie de la vie. 2024-01-16 10:21:20,273 - DEBUG - joeynmt.training - Tokenized source: ['▁All', '▁of', '▁these', '▁people', 's', '▁teach', '▁us', '▁that', '▁there', '▁are', '▁other', '▁ways', '▁of', '▁being', ',', '▁other', '▁ways', '▁of', '▁thinking', ',', '▁other', '▁ways', '▁of', '▁orient', 'ing', '▁yourself', '▁in', '▁the', '▁Earth', '.'] 2024-01-16 10:21:20,273 - DEBUG - joeynmt.training - Tokenized reference: ['▁Tous', '▁ces', '▁peuple', 's', '▁nous', '▁enseigne', 'nt', '▁qu', "'", 'il', '▁y', '▁a', '▁d', "'", 'autres', '▁façons', '▁d', "'", 'être', ',', '▁d', "'", 'autres', '▁façons', '▁de', '▁penser', ',', '▁d', "'", 'autres', '▁manières', '▁de', '▁s', "'", 'orient', 'er', '▁sur', '▁Terre', '.'] 2024-01-16 10:21:20,273 - DEBUG - joeynmt.training - Tokenized hypothesis: ['', '▁Et', '▁que', '▁ce', '▁soit', '▁le', '▁Pen', 'an', '▁dans', '▁les', '▁forêts', '▁du', '▁Bo', 'r', 'né', 'o', ',', '▁ou', '▁les', '▁a', 'co', 'ly', 'tes', '▁V', 'oodoo', '▁à', '▁Haïti', ',', '▁ou', '▁bien', '▁les', '▁guerrier', 's', '▁dans', '▁le', '▁désert', '▁du', '▁K', 'ais', 'ut', '▁au', '▁nord', '▁du', '▁Kenya', ',', '▁le', '▁C', 'ur', 'ende', 'ro', '▁dans', '▁les', '▁montagnes', '▁des', '▁And', 'es', ',', '▁ou', '▁bien', '▁un', '▁car', 'ava', 'n', 's', 'é', 'ra', 'il', '▁en', '▁plein', '▁milieu', '▁du', '▁Sahara', '.', '▁A', '▁propos', ',', '▁c', "'", 'est', '▁la', '▁personne', '▁avec', '▁qui', '▁j', "'", 'ai', '▁voyagé', '▁dans', '▁le', '▁désert', '▁il', '▁y', '▁un', '▁mois', ',', '▁ou', '▁effectivement', ',', '▁le', '▁gardien', '▁de', '▁troupeau', '▁de', '▁Yak', 's', '▁sur', '▁les', '▁fla', 'n', 'c', 's', '▁du', '▁Q', 'o', 'mo', 'lang', 'ma', ',', '▁l', "'", 'Everest', ',', '▁la', '▁dé', 'esse', '▁du', '▁monde', '.', '', '▁Et', '▁il', '▁y', '▁a', '▁été', '▁un', '▁monde', ',', '▁et', '▁la', '▁vie', ',', '▁et', '▁la', '▁vie', ',', '▁la', '▁vie', ',', '▁la', '▁vie', ',', '▁la', '▁vie', '▁est', '▁un', '▁monde', ',', '▁et', '▁la', '▁vie', '▁de', '▁la', '▁vie', '▁de', '▁la', '▁vie', '▁de', '▁la', '▁vie', '▁de', '▁la', '▁vie', '▁de', '▁la', '▁vie', '▁de', '▁la', '▁vie', '▁de', '▁la', '▁vie', '.'] 2024-01-16 10:21:20,274 - INFO - joeynmt.training - Example #3 2024-01-16 10:21:20,274 - INFO - joeynmt.training - Source: It's the symbol of all that we are and all that we can be as an astonishingly inquisitive species. 2024-01-16 10:21:20,274 - INFO - joeynmt.training - Reference: C'est le symbole de tout ce que nous sommes et tout ce que nous pouvons être en tant qu'espèce dotée d'une curiosité stupéfiante. 2024-01-16 10:21:20,274 - INFO - joeynmt.training - Hypothesis: Et nous avons pas que nous avons pas de la vie, et nous avons pas de la vie, et nous avons pas de la vie. 2024-01-16 10:21:20,275 - DEBUG - joeynmt.training - Tokenized source: ['▁It', "'", 's', '▁the', '▁symbol', '▁of', '▁all', '▁that', '▁we', '▁are', '▁and', '▁all', '▁that', '▁we', '▁can', '▁be', '▁as', '▁an', '▁astonishing', 'ly', '▁in', 'qui', 's', 'it', 'ive', '▁species', '.'] 2024-01-16 10:21:20,275 - DEBUG - joeynmt.training - Tokenized reference: ['▁C', "'", 'est', '▁le', '▁symbole', '▁de', '▁tout', '▁ce', '▁que', '▁nous', '▁sommes', '▁et', '▁tout', '▁ce', '▁que', '▁nous', '▁pouvons', '▁être', '▁en', '▁tant', '▁qu', "'", 'espèce', '▁do', 't', 'ée', '▁d', "'", 'une', '▁curiosité', '▁stupéfiant', 'e', '.'] 2024-01-16 10:21:20,275 - DEBUG - joeynmt.training - Tokenized hypothesis: ['', '▁L', "'", 'e', 'th', 'no', 's', 'ph', 'ère', '▁est', '▁l', "'", 'héritage', '▁de', '▁l', "'", 'humanité', '.', '', '▁Et', '▁nous', '▁avons', '▁pas', '▁que', '▁nous', '▁avons', '▁pas', '▁de', '▁la', '▁vie', ',', '▁et', '▁nous', '▁avons', '▁pas', '▁de', '▁la', '▁vie', ',', '▁et', '▁nous', '▁avons', '▁pas', '▁de', '▁la', '▁vie', '.'] 2024-01-16 10:22:37,445 - INFO - joeynmt.training - Epoch 1, Step: 2100, Batch Loss: 4.285705, Batch Acc: 0.246956, Tokens per Sec: 16146, Lr: 0.000042 2024-01-16 10:23:53,699 - INFO - joeynmt.training - Epoch 1, Step: 2200, Batch Loss: 4.225679, Batch Acc: 0.253556, Tokens per Sec: 16449, Lr: 0.000044 2024-01-16 10:25:10,017 - INFO - joeynmt.training - Epoch 1, Step: 2300, Batch Loss: 4.194219, Batch Acc: 0.260617, Tokens per Sec: 16288, Lr: 0.000046 2024-01-16 10:26:26,606 - INFO - joeynmt.training - Epoch 1, Step: 2400, Batch Loss: 4.081372, Batch Acc: 0.269474, Tokens per Sec: 16289, Lr: 0.000048 2024-01-16 10:27:42,873 - INFO - joeynmt.training - Epoch 1, Step: 2500, Batch Loss: 4.021136, Batch Acc: 0.277450, Tokens per Sec: 16370, Lr: 0.000050 2024-01-16 10:28:59,046 - INFO - joeynmt.training - Epoch 1, Step: 2600, Batch Loss: 3.986295, Batch Acc: 0.286659, Tokens per Sec: 16359, Lr: 0.000052 2024-01-16 10:30:16,034 - INFO - joeynmt.training - Epoch 1, Step: 2700, Batch Loss: 3.978336, Batch Acc: 0.295439, Tokens per Sec: 16265, Lr: 0.000054 2024-01-16 10:30:48,449 - INFO - joeynmt.training - Epoch 1, total training loss: 14305.52, num. of seqs: 702202, num. of tokens: 34266040, 2093.6051[sec] 2024-01-16 10:30:48,459 - INFO - joeynmt.training - EPOCH 2 2024-01-16 10:31:32,246 - INFO - joeynmt.training - Epoch 2, Step: 2800, Batch Loss: 3.849685, Batch Acc: 0.307792, Tokens per Sec: 16191, Lr: 0.000056 2024-01-16 10:32:49,076 - INFO - joeynmt.training - Epoch 2, Step: 2900, Batch Loss: 3.754404, Batch Acc: 0.312482, Tokens per Sec: 16316, Lr: 0.000058 2024-01-16 10:34:06,246 - INFO - joeynmt.training - Epoch 2, Step: 3000, Batch Loss: 3.754129, Batch Acc: 0.321229, Tokens per Sec: 16275, Lr: 0.000060 2024-01-16 10:34:06,248 - INFO - joeynmt.datasets - Sample random subset from dev data: n=500, seed=3042 2024-01-16 10:34:06,248 - INFO - joeynmt.prediction - Predicting 500 example(s)... (Greedy decoding with min_output_length=1, max_output_length=512, return_prob='none', generate_unk=True, repetition_penalty=-1, no_repeat_ngram_size=-1) 2024-01-16 10:35:23,237 - INFO - joeynmt.prediction - Generation took 76.9812[sec]. 2024-01-16 10:35:23,394 - INFO - joeynmt.metrics - nrefs:1|case:lc|eff:no|tok:13a|smooth:exp|version:2.4.0 2024-01-16 10:35:23,394 - INFO - joeynmt.prediction - Evaluation result (greedy): bleu: 2.02, loss: 4.21, ppl: 67.30, acc: 0.26, 0.1305[sec] 2024-01-16 10:35:23,395 - INFO - joeynmt.training - Hooray! New best validation result [bleu]! 2024-01-16 10:35:26,272 - INFO - joeynmt.training - Checkpoint saved in models/iwslt14_prompt/3000.ckpt. 2024-01-16 10:35:26,275 - INFO - joeynmt.training - Example #0 2024-01-16 10:35:26,276 - INFO - joeynmt.training - Source: You know, one of the intense pleasures of travel and one of the delights of ethnographic research is the opportunity to live amongst those who have not forgotten the old ways, who still feel their past in the wind, touch it in stones polished by rain, taste it in the bitter leaves of plants. 2024-01-16 10:35:26,276 - INFO - joeynmt.training - Reference: Vous savez, un des plaisirs intenses du voyage et un des délices de la recherche ethnographique est la possibilité de vivre parmi ceux qui n'ont pas oublié les anciennes coutumes, qui ressentent encore leur passé souffler dans le vent, qui le touchent dans les pierres polies par la pluie, le dégustent dans les feuilles amères des plantes. 2024-01-16 10:35:26,276 - INFO - joeynmt.training - Hypothesis: Vous savez, une sorte de la plupart de la vie de la vie et de la la la la la la la vie est que les gens qui ne sont pas de la vie qui ne sont pas de ces gens qui ont pas de la vie, qui ont été été été été été en train de la vie, qui ont été été en train de la vie, dans le monde, dans le monde, en fait le monde de la vie de la vie. 2024-01-16 10:35:26,277 - DEBUG - joeynmt.training - Tokenized source: ['▁You', '▁know', ',', '▁one', '▁of', '▁the', '▁intense', '▁pleasure', 's', '▁of', '▁travel', '▁and', '▁one', '▁of', '▁the', '▁de', 'light', 's', '▁of', '▁et', 'h', 'n', 'ographic', '▁research', '▁is', '▁the', '▁opportunity', '▁to', '▁live', '▁among', 'st', '▁those', '▁who', '▁have', '▁not', '▁forgotten', '▁the', '▁old', '▁ways', ',', '▁who', '▁still', '▁feel', '▁their', '▁past', '▁in', '▁the', '▁wind', ',', '▁touch', '▁it', '▁in', '▁stones', '▁polish', 'ed', '▁by', '▁rain', ',', '▁taste', '▁it', '▁in', '▁the', '▁bitter', '▁leaves', '▁of', '▁plants', '.'] 2024-01-16 10:35:26,277 - DEBUG - joeynmt.training - Tokenized reference: ['▁Vous', '▁savez', ',', '▁un', '▁des', '▁plaisir', 's', '▁intense', 's', '▁du', '▁voyage', '▁et', '▁un', '▁des', '▁dé', 'lic', 'es', '▁de', '▁la', '▁recherche', '▁et', 'h', 'no', 'graph', 'ique', '▁est', '▁la', '▁possibilité', '▁de', '▁vivre', '▁parmi', '▁ceux', '▁qui', '▁n', "'", 'ont', '▁pas', '▁oublié', '▁les', '▁anciennes', '▁coutume', 's', ',', '▁qui', '▁ressentent', '▁encore', '▁leur', '▁passé', '▁souffle', 'r', '▁dans', '▁le', '▁vent', ',', '▁qui', '▁le', '▁touchent', '▁dans', '▁les', '▁pierres', '▁poli', 'es', '▁par', '▁la', '▁pluie', ',', '▁le', '▁dé', 'gu', 'sten', 't', '▁dans', '▁les', '▁feuilles', '▁a', 'mère', 's', '▁des', '▁plantes', '.'] 2024-01-16 10:35:26,277 - DEBUG - joeynmt.training - Tokenized hypothesis: ['', '', '▁Vous', '▁savez', ',', '▁une', '▁sorte', '▁de', '▁la', '▁plupart', '▁de', '▁la', '▁vie', '▁de', '▁la', '▁vie', '▁et', '▁de', '▁la', '▁la', '▁la', '▁la', '▁la', '▁la', '▁la', '▁vie', '▁est', '▁que', '▁les', '▁gens', '▁qui', '▁ne', '▁sont', '▁pas', '▁de', '▁la', '▁vie', '▁qui', '▁ne', '▁sont', '▁pas', '▁de', '▁ces', '▁gens', '▁qui', '▁ont', '▁pas', '▁de', '▁la', '▁vie', ',', '▁qui', '▁ont', '▁été', '▁été', '▁été', '▁été', '▁été', '▁en', '▁train', '▁de', '▁la', '▁vie', ',', '▁qui', '▁ont', '▁été', '▁été', '▁en', '▁train', '▁de', '▁la', '▁vie', ',', '▁dans', '▁le', '▁monde', ',', '▁dans', '▁le', '▁monde', ',', '▁en', '▁fait', '▁le', '▁monde', '▁de', '▁la', '▁vie', '▁de', '▁la', '▁vie', '.'] 2024-01-16 10:35:26,277 - INFO - joeynmt.training - Example #1 2024-01-16 10:35:26,278 - INFO - joeynmt.training - Source: Nous sommes tous nés. Nous apportons nos enfants dans ce monde. 2024-01-16 10:35:26,278 - INFO - joeynmt.training - Reference: We're all born. We all bring our children into the world. 2024-01-16 10:35:26,278 - INFO - joeynmt.training - Hypothesis: We're all all at the people. We're going to go to the world in the world. 2024-01-16 10:35:26,278 - DEBUG - joeynmt.training - Tokenized source: ['▁Nous', '▁sommes', '▁tous', '▁né', 's', '.', '▁Nous', '▁apport', 'ons', '▁nos', '▁enfants', '▁dans', '▁ce', '▁monde', '.'] 2024-01-16 10:35:26,279 - DEBUG - joeynmt.training - Tokenized reference: ['▁We', "'", 're', '▁all', '▁born', '.', '▁We', '▁all', '▁bring', '▁our', '▁children', '▁into', '▁the', '▁world', '.'] 2024-01-16 10:35:26,279 - DEBUG - joeynmt.training - Tokenized hypothesis: ['', '▁And', '▁of', '▁course', ',', '▁we', '▁all', '▁share', '▁the', '▁same', '▁adaptive', '▁imperative', 's', '.', '', '▁We', "'", 're', '▁all', '▁all', '▁at', '▁the', '▁people', '.', '▁We', "'", 're', '▁going', '▁to', '▁go', '▁to', '▁the', '▁world', '▁in', '▁the', '▁world', '.'] 2024-01-16 10:35:26,279 - INFO - joeynmt.training - Example #2 2024-01-16 10:35:26,279 - INFO - joeynmt.training - Source: Aujourd'hui, les innombrables cultures dans le monde constituent un tissu de vie spirituelle et culturelle qui enveloppe la planète, et qui est aussi important pour le bien-être de la planète que l'est également le tissu biologique de la vie que vous connaissez en tant que biosphère. 2024-01-16 10:35:26,279 - INFO - joeynmt.training - Reference: Now, together the myriad cultures of the world make up a web of spiritual life and cultural life that envelops the planet, and is as important to the well-being of the planet as indeed is the biological web of life that you know as a biosphere. 2024-01-16 10:35:26,279 - INFO - joeynmt.training - Hypothesis: So, the most of the world, the world in the world in the world, and the world is a very important way that is that the world is that the world is that the world is that the same thing that you're going to be able to be able to be able to be able to be able to be in the same time. 2024-01-16 10:35:26,280 - DEBUG - joeynmt.training - Tokenized source: ['▁A', 'ujourd', "'", 'hui', ',', '▁les', '▁', 'innombrables', '▁cultures', '▁dans', '▁le', '▁monde', '▁constituent', '▁un', '▁tissu', '▁de', '▁vie', '▁spirituelle', '▁et', '▁culturelle', '▁qui', '▁enveloppe', '▁la', '▁planète', ',', '▁et', '▁qui', '▁est', '▁aussi', '▁important', '▁pour', '▁le', '▁bien', '-', 'être', '▁de', '▁la', '▁planète', '▁que', '▁l', "'", 'est', '▁également', '▁le', '▁tissu', '▁biologique', '▁de', '▁la', '▁vie', '▁que', '▁vous', '▁connaissez', '▁en', '▁tant', '▁que', '▁biosphère', '.'] 2024-01-16 10:35:26,280 - DEBUG - joeynmt.training - Tokenized reference: ['▁Now', ',', '▁together', '▁the', '▁myriad', '▁cultures', '▁of', '▁the', '▁world', '▁make', '▁up', '▁a', '▁web', '▁of', '▁spiritual', '▁life', '▁and', '▁cultural', '▁life', '▁that', '▁en', 've', 'lop', 's', '▁the', '▁planet', ',', '▁and', '▁is', '▁as', '▁important', '▁to', '▁the', '▁well', '-', 'being', '▁of', '▁the', '▁planet', '▁as', '▁indeed', '▁is', '▁the', '▁biological', '▁web', '▁of', '▁life', '▁that', '▁you', '▁know', '▁as', '▁a', '▁biosphere', '.'] 2024-01-16 10:35:26,280 - DEBUG - joeynmt.training - Tokenized hypothesis: ['', '▁And', '▁this', '▁is', '▁an', '▁idea', ',', '▁if', '▁you', '▁think', '▁about', '▁it', ',', '▁can', '▁only', '▁fill', '▁you', '▁with', '▁hope', '.', '', '▁So', ',', '▁the', '▁most', '▁of', '▁the', '▁world', ',', '▁the', '▁world', '▁in', '▁the', '▁world', '▁in', '▁the', '▁world', ',', '▁and', '▁the', '▁world', '▁is', '▁a', '▁very', '▁important', '▁way', '▁that', '▁is', '▁that', '▁the', '▁world', '▁is', '▁that', '▁the', '▁world', '▁is', '▁that', '▁the', '▁world', '▁is', '▁that', '▁the', '▁same', '▁thing', '▁that', '▁you', "'", 're', '▁going', '▁to', '▁be', '▁able', '▁to', '▁be', '▁able', '▁to', '▁be', '▁able', '▁to', '▁be', '▁able', '▁to', '▁be', '▁able', '▁to', '▁be', '▁in', '▁the', '▁same', '▁time', '.'] 2024-01-16 10:35:26,280 - INFO - joeynmt.training - Example #3 2024-01-16 10:35:26,281 - INFO - joeynmt.training - Source: It's the symbol of all that we are and all that we can be as an astonishingly inquisitive species. 2024-01-16 10:35:26,281 - INFO - joeynmt.training - Reference: C'est le symbole de tout ce que nous sommes et tout ce que nous pouvons être en tant qu'espèce dotée d'une curiosité stupéfiante. 2024-01-16 10:35:26,281 - INFO - joeynmt.training - Hypothesis: C'est la la plupart de la vie de nous nous nous nous pouvons être un peu de la plus de la plus de la plus de la plus de la plus de la plus de la plus de la plus. 2024-01-16 10:35:26,281 - DEBUG - joeynmt.training - Tokenized source: ['▁It', "'", 's', '▁the', '▁symbol', '▁of', '▁all', '▁that', '▁we', '▁are', '▁and', '▁all', '▁that', '▁we', '▁can', '▁be', '▁as', '▁an', '▁astonishing', 'ly', '▁in', 'qui', 's', 'it', 'ive', '▁species', '.'] 2024-01-16 10:35:26,282 - DEBUG - joeynmt.training - Tokenized reference: ['▁C', "'", 'est', '▁le', '▁symbole', '▁de', '▁tout', '▁ce', '▁que', '▁nous', '▁sommes', '▁et', '▁tout', '▁ce', '▁que', '▁nous', '▁pouvons', '▁être', '▁en', '▁tant', '▁qu', "'", 'espèce', '▁do', 't', 'ée', '▁d', "'", 'une', '▁curiosité', '▁stupéfiant', 'e', '.'] 2024-01-16 10:35:26,282 - DEBUG - joeynmt.training - Tokenized hypothesis: ['', '▁L', "'", 'e', 'th', 'no', 's', 'ph', 'ère', '▁est', '▁l', "'", 'héritage', '▁de', '▁l', "'", 'humanité', '.', '', '▁C', "'", 'est', '▁la', '▁la', '▁plupart', '▁de', '▁la', '▁vie', '▁de', '▁nous', '▁nous', '▁nous', '▁nous', '▁pouvons', '▁être', '▁un', '▁peu', '▁de', '▁la', '▁plus', '▁de', '▁la', '▁plus', '▁de', '▁la', '▁plus', '▁de', '▁la', '▁plus', '▁de', '▁la', '▁plus', '▁de', '▁la', '▁plus', '▁de', '▁la', '▁plus', '▁de', '▁la', '▁plus', '.'] 2024-01-16 10:36:42,913 - INFO - joeynmt.training - Epoch 2, Step: 3100, Batch Loss: 3.664998, Batch Acc: 0.330314, Tokens per Sec: 16257, Lr: 0.000062 2024-01-16 10:37:59,267 - INFO - joeynmt.training - Epoch 2, Step: 3200, Batch Loss: 3.566615, Batch Acc: 0.338621, Tokens per Sec: 16349, Lr: 0.000064 2024-01-16 10:39:16,055 - INFO - joeynmt.training - Epoch 2, Step: 3300, Batch Loss: 3.484769, Batch Acc: 0.347524, Tokens per Sec: 16325, Lr: 0.000066 2024-01-16 10:40:32,038 - INFO - joeynmt.training - Epoch 2, Step: 3400, Batch Loss: 3.486224, Batch Acc: 0.356624, Tokens per Sec: 16420, Lr: 0.000068 2024-01-16 10:41:47,712 - INFO - joeynmt.training - Epoch 2, Step: 3500, Batch Loss: 3.415474, Batch Acc: 0.367184, Tokens per Sec: 16425, Lr: 0.000070 2024-01-16 10:43:03,903 - INFO - joeynmt.training - Epoch 2, Step: 3600, Batch Loss: 3.294894, Batch Acc: 0.374602, Tokens per Sec: 16432, Lr: 0.000072 2024-01-16 10:44:19,764 - INFO - joeynmt.training - Epoch 2, Step: 3700, Batch Loss: 3.311182, Batch Acc: 0.384667, Tokens per Sec: 16447, Lr: 0.000074 2024-01-16 10:45:40,969 - INFO - joeynmt.training - Epoch 2, Step: 3800, Batch Loss: 3.239654, Batch Acc: 0.393206, Tokens per Sec: 15478, Lr: 0.000076 2024-01-16 10:46:57,545 - INFO - joeynmt.training - Epoch 2, Step: 3900, Batch Loss: 3.148585, Batch Acc: 0.404261, Tokens per Sec: 16381, Lr: 0.000078 2024-01-16 10:48:14,022 - INFO - joeynmt.training - Epoch 2, Step: 4000, Batch Loss: 3.106784, Batch Acc: 0.410737, Tokens per Sec: 16388, Lr: 0.000080 2024-01-16 10:48:14,089 - INFO - joeynmt.datasets - Sample random subset from dev data: n=500, seed=4042 2024-01-16 10:48:14,112 - INFO - joeynmt.prediction - Predicting 500 example(s)... (Greedy decoding with min_output_length=1, max_output_length=512, return_prob='none', generate_unk=True, repetition_penalty=-1, no_repeat_ngram_size=-1) 2024-01-16 10:50:15,719 - INFO - joeynmt.prediction - Generation took 121.5720[sec]. 2024-01-16 10:50:15,931 - INFO - joeynmt.metrics - nrefs:1|case:lc|eff:no|tok:13a|smooth:exp|version:2.4.0 2024-01-16 10:50:15,931 - INFO - joeynmt.prediction - Evaluation result (greedy): bleu: 6.85, loss: 3.51, ppl: 33.34, acc: 0.37, 0.0666[sec] 2024-01-16 10:50:15,932 - INFO - joeynmt.training - Hooray! New best validation result [bleu]! 2024-01-16 10:50:19,092 - INFO - joeynmt.training - Checkpoint saved in models/iwslt14_prompt/4000.ckpt. 2024-01-16 10:50:19,097 - INFO - joeynmt.training - Example #0 2024-01-16 10:50:19,097 - INFO - joeynmt.training - Source: C'est le symbole de tout ce que nous sommes et tout ce que nous pouvons être en tant qu'espèce dotée d'une curiosité stupéfiante. 2024-01-16 10:50:19,097 - INFO - joeynmt.training - Reference: It's the symbol of all that we are and all that we can be as an astonishingly inquisitive species. 2024-01-16 10:50:19,098 - INFO - joeynmt.training - Hypothesis: It's the process of all that we're and all that we can be in as a lot of the politically-termly. 2024-01-16 10:50:19,098 - DEBUG - joeynmt.training - Tokenized source: ['▁C', "'", 'est', '▁le', '▁symbole', '▁de', '▁tout', '▁ce', '▁que', '▁nous', '▁sommes', '▁et', '▁tout', '▁ce', '▁que', '▁nous', '▁pouvons', '▁être', '▁en', '▁tant', '▁qu', "'", 'espèce', '▁do', 't', 'ée', '▁d', "'", 'une', '▁curiosité', '▁stupéfiant', 'e', '.'] 2024-01-16 10:50:19,098 - DEBUG - joeynmt.training - Tokenized reference: ['▁It', "'", 's', '▁the', '▁symbol', '▁of', '▁all', '▁that', '▁we', '▁are', '▁and', '▁all', '▁that', '▁we', '▁can', '▁be', '▁as', '▁an', '▁astonishing', 'ly', '▁in', 'qui', 's', 'it', 'ive', '▁species', '.'] 2024-01-16 10:50:19,099 - DEBUG - joeynmt.training - Tokenized hypothesis: ['', '▁The', '▁et', 'h', 'n', 'osphere', '▁is', '▁humanity', "'", 's', '▁great', '▁legacy', '.', '', '▁It', "'", 's', '▁the', '▁process', '▁of', '▁all', '▁that', '▁we', "'", 're', '▁and', '▁all', '▁that', '▁we', '▁can', '▁be', '▁in', '▁as', '▁a', '▁lot', '▁of', '▁the', '▁political', 'ly', '-', 'term', 'ly', '.'] 2024-01-16 10:50:19,099 - INFO - joeynmt.training - Example #1 2024-01-16 10:50:19,099 - INFO - joeynmt.training - Source: Enfin, ce que je voudrais faire avec vous aujourd'hui c'est vous emmener, en quelque sorte, faire un voyage dans l'ethnosphère -- un court voyage dans l'ethnosphère pour tenter de vous expliquer, en fait, ce qui se perd. 2024-01-16 10:50:19,099 - INFO - joeynmt.training - Reference: And so, what I'd like to do with you today is sort of take you on a journey through the ethnosphere, a brief journey through the ethnosphere, to try to begin to give you a sense of what in fact is being lost. 2024-01-16 10:50:19,099 - INFO - joeynmt.training - Hypothesis: And then, what I want to do with you today is you to go, to you're going to go, to a kind of work in the 'Eccyyy, a long-year-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a--------------------a-a-a-a-----------------------------------------a-a--------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------- 2024-01-16 10:50:19,100 - DEBUG - joeynmt.training - Tokenized source: ['▁Enfin', ',', '▁ce', '▁que', '▁je', '▁voudrais', '▁faire', '▁avec', '▁vous', '▁aujourd', "'", 'hui', '▁c', "'", 'est', '▁vous', '▁emmener', ',', '▁en', '▁quelque', '▁sorte', ',', '▁faire', '▁un', '▁voyage', '▁dans', '▁l', "'", 'e', 'th', 'no', 's', 'ph', 'ère', '▁--', '▁un', '▁court', '▁voyage', '▁dans', '▁l', "'", 'e', 'th', 'no', 's', 'ph', 'ère', '▁pour', '▁tenter', '▁de', '▁vous', '▁expliquer', ',', '▁en', '▁fait', ',', '▁ce', '▁qui', '▁se', '▁perd', '.'] 2024-01-16 10:50:19,100 - DEBUG - joeynmt.training - Tokenized reference: ['▁And', '▁so', ',', '▁what', '▁I', "'", 'd', '▁like', '▁to', '▁do', '▁with', '▁you', '▁today', '▁is', '▁sort', '▁of', '▁take', '▁you', '▁on', '▁a', '▁journey', '▁through', '▁the', '▁et', 'h', 'n', 'osphere', ',', '▁a', '▁brief', '▁journey', '▁through', '▁the', '▁et', 'h', 'n', 'osphere', ',', '▁to', '▁try', '▁to', '▁begin', '▁to', '▁give', '▁you', '▁a', '▁sense', '▁of', '▁what', '▁in', '▁fact', '▁is', '▁being', '▁lost', '.'] 2024-01-16 10:50:19,100 - DEBUG - joeynmt.training - Tokenized hypothesis: ['', '▁And', '▁you', "'", 'll', '▁suddenly', '▁discover', '▁what', '▁it', '▁would', '▁be', '▁like', '▁to', '▁be', '▁unable', '▁to', '▁speak', '▁your', '▁own', '▁language', '.', '', '▁And', '▁then', ',', '▁what', '▁I', '▁want', '▁to', '▁do', '▁with', '▁you', '▁today', '▁is', '▁you', '▁to', '▁go', ',', '▁to', '▁you', "'", 're', '▁going', '▁to', '▁go', ',', '▁to', '▁a', '▁kind', '▁of', '▁work', '▁in', '▁the', '▁', "'", 'E', 'c', 'c', 'y', 'y', 'y', ',', '▁a', '▁long', '-', 'year', '-', 'a', '-', 'a', '-', 'a', '-', 'a', '-', 'a', '-', 'a', '-', 'a', '-', 'a', '-', 'a', '-', 'a', '-', 'a', '-', 'a', '-', 'a', '-', 'a', '-', 'a', '-', 'a', '-', 'a', '-', 'a', '-', 'a', '-', 'a', '-', 'a', '-', 'a', '-', 'a', '-', 'a', '-', 'a', '-', 'a', '-', 'a', '-', '-', '-', '-', '-', '-', '-', '-', '-', '-', '-', '-', '-', '-', '-', '-', '-', '-', '-', '-', 'a', '-', 'a', '-', 'a', '-', 'a', '-', '-', '-', '-', '-', '-', '-', '-', '-', '-', '-', '-', '-', '-', '-', '-', '-', '-', '-', '-', '-', '-', '-', '-', '-', '-', '-', '-', '-', '-', '-', '-', '-', '-', '-', '-', '-', '-', '-', '-', '-', 'a', '-', 'a', '-', '-', '-', '-', '-', '-', '-', '-', '-', '-', '-', '-', '-', '-', '-', '-', '-', '-', '-', '-', '-', '-', '-', '-', '-', '-', '-', '-', '-', '-', '-', '-', '-', '-', '-', '-', '-', '-', '-', '-', '-', '-', '-', '-', '-', '-', '-', '-', '-', '-', '-', '-', '-', '-', '-', '-', '-', '-', '-', '-', '-', '-', '-', '-', '-', '-', '-', '-', '-', '-', '-', '-', '-', '-', '-', '-', '-', '-', '-', '-', '-', '-', '-', '-', '-', '-', '-', '-', '-', '-', '-', '-', '-', '-', '-', '-', '-', '-', '-', '-', '-', '-', '-', '-', '-', '-', '-', '-', '-', '-', '-', '-', '-', '-', '-', '-', '-', '-', '-', '-', '-', '-', '-', '-', '-', '-', '-', '-', '-', '-', '-', '-', '-', '-', '-', '-', '-', '-', '-', '-', '-', '-', '-', '-', '-', '-', '-', '-', '-', '-', '-', '-', '-', '-', '-', '-', '-', '-', '-', '-', '-', '-', '-', '-', '-', '-', '-', '-', '-', '-', '-', '-', '-', '-', '-', '-', '-', '-', '-', '-', '-', '-', '-', '-', '-', '-', '-', '-', '-', '-', '-', '-', '-', '-', '-', '-', '-', '-', '-', '-', '-', '-', '-', '-', '-', '-', '-', '-', '-', '-', '-', '-', '-', '-', '-', '-', '-', '-', '-', '-', '-', '-', '-', '-', '-', '-', '-', '-', '-', '-', '-', '-', '-', '-', '-', '-', '-', '-', '-', '-', '-', '-', '-', '-', '-', '-', '-', '-', '-', '-', '-', '-', '-', '-', '-', '-', '-', '-', '-', '-', '-', '-', '-', '-', '-', '-', '-', '-', '-', '-', '-', '-', '-', '-', '-', '-', '-', '-', '-', '-', '-', '-', '-', '-', '-', '-', '-', '-', '-', '-', '-', '-', '-', '-', '-', '-', '-', '-', '-', '-', '-', '-', '-', '-', '-', '-', '-', '-', '-', '-', '-', '-', '-', '-', '-', '-', '-', '-', '-', '-'] 2024-01-16 10:50:19,100 - INFO - joeynmt.training - Example #2 2024-01-16 10:50:19,101 - INFO - joeynmt.training - Source: Ils écoutent tout simplement et puis commencent à parler. 2024-01-16 10:50:19,101 - INFO - joeynmt.training - Reference: They simply listen and then begin to speak. 2024-01-16 10:50:19,101 - INFO - joeynmt.training - Hypothesis: They're just going to be able to talk and then start talking about it. 2024-01-16 10:50:19,101 - DEBUG - joeynmt.training - Tokenized source: ['▁Ils', '▁écoute', 'nt', '▁tout', '▁simplement', '▁et', '▁puis', '▁commencent', '▁à', '▁parler', '.'] 2024-01-16 10:50:19,101 - DEBUG - joeynmt.training - Tokenized reference: ['▁They', '▁simply', '▁listen', '▁and', '▁then', '▁begin', '▁to', '▁speak', '.'] 2024-01-16 10:50:19,101 - DEBUG - joeynmt.training - Tokenized hypothesis: ['', '▁And', '▁this', '▁is', '▁all', '▁rooted', '▁in', '▁the', '▁myth', 'ological', '▁past', ',', '▁yet', '▁the', '▁curious', '▁thing', '▁is', '▁in', '▁these', '▁long', '▁houses', ',', '▁where', '▁there', '▁are', '▁six', '▁or', '▁seven', '▁languages', '▁spoken', '▁because', '▁of', '▁in', 'term', 'arri', 'age', ',', '▁you', '▁never', '▁hear', '▁anyone', '▁practicing', '▁a', '▁language', '.', '', '▁They', "'", 're', '▁just', '▁going', '▁to', '▁be', '▁able', '▁to', '▁talk', '▁and', '▁then', '▁start', '▁talking', '▁about', '▁it', '.'] 2024-01-16 10:50:19,101 - INFO - joeynmt.training - Example #3 2024-01-16 10:50:19,102 - INFO - joeynmt.training - Source: Ils ont fait tomber du ciel des photos brillantes, d'eux-mêmes, 8 par 10, ce que l'on pourrait qualifier de témoignage d'amitié, oubliant que ces peuples des forêts tropicales n'avaient jamais rien vu en 2 dimensions de leur vie. 2024-01-16 10:50:19,102 - INFO - joeynmt.training - Reference: They dropped from the air 8 x 10 glossy photographs of themselves in what we would say to be friendly gestures, forgetting that these people of the rainforest had never seen anything two-dimensional in their lives. 2024-01-16 10:50:19,102 - INFO - joeynmt.training - Hypothesis: They've got to go to the sky of the nights, of them, five, five, which we could be able to be able to be able to be ad-up-s-up, to be able to be the people who were never seen in two-dimensional life. 2024-01-16 10:50:19,103 - DEBUG - joeynmt.training - Tokenized source: ['▁Ils', '▁ont', '▁fait', '▁tomber', '▁du', '▁ciel', '▁des', '▁photos', '▁brillante', 's', ',', '▁d', "'", 'eux', '-', 'mêmes', ',', '▁8', '▁par', '▁10', ',', '▁ce', '▁que', '▁l', "'", 'on', '▁pourrait', '▁qualifi', 'er', '▁de', '▁témoignage', '▁d', "'", 'amitié', ',', '▁', 'oubli', 'ant', '▁que', '▁ces', '▁peuple', 's', '▁des', '▁forêts', '▁tropical', 'es', '▁n', "'", 'avaient', '▁jamais', '▁rien', '▁vu', '▁en', '▁2', '▁dimensions', '▁de', '▁leur', '▁vie', '.'] 2024-01-16 10:50:19,103 - DEBUG - joeynmt.training - Tokenized reference: ['▁They', '▁dropped', '▁from', '▁the', '▁air', '▁8', '▁x', '▁10', '▁g', 'los', 's', 'y', '▁photographs', '▁of', '▁themselves', '▁in', '▁what', '▁we', '▁would', '▁say', '▁to', '▁be', '▁friendly', '▁gestures', ',', '▁forget', 'ting', '▁that', '▁these', '▁people', '▁of', '▁the', '▁rainforest', '▁had', '▁never', '▁seen', '▁anything', '▁two', '-', 'dimensional', '▁in', '▁their', '▁lives', '.'] 2024-01-16 10:50:19,103 - DEBUG - joeynmt.training - Tokenized hypothesis: ['', '▁In', '▁1957', ',', '▁five', '▁mission', 'aries', '▁attempt', 'ed', '▁contact', '▁and', '▁made', '▁a', '▁critical', '▁mistake', '.', '', '▁They', "'", 've', '▁got', '▁to', '▁go', '▁to', '▁the', '▁sky', '▁of', '▁the', '▁night', 's', ',', '▁of', '▁them', ',', '▁five', ',', '▁five', ',', '▁which', '▁we', '▁could', '▁be', '▁able', '▁to', '▁be', '▁able', '▁to', '▁be', '▁able', '▁to', '▁be', '▁a', 'd', '-', 'up', '-', 's', '-', 'up', ',', '▁to', '▁be', '▁able', '▁to', '▁be', '▁the', '▁people', '▁who', '▁were', '▁never', '▁seen', '▁in', '▁two', '-', 'dimensional', '▁life', '.'] 2024-01-16 10:51:35,123 - INFO - joeynmt.training - Epoch 2, Step: 4100, Batch Loss: 3.097340, Batch Acc: 0.420527, Tokens per Sec: 16447, Lr: 0.000082 2024-01-16 10:52:51,921 - INFO - joeynmt.training - Epoch 2, Step: 4200, Batch Loss: 3.088614, Batch Acc: 0.428939, Tokens per Sec: 16300, Lr: 0.000084 2024-01-16 10:54:07,334 - INFO - joeynmt.training - Epoch 2, Step: 4300, Batch Loss: 2.923417, Batch Acc: 0.436478, Tokens per Sec: 16544, Lr: 0.000086 2024-01-16 10:55:23,805 - INFO - joeynmt.training - Epoch 2, Step: 4400, Batch Loss: 2.957743, Batch Acc: 0.445010, Tokens per Sec: 16351, Lr: 0.000088 2024-01-16 10:56:39,577 - INFO - joeynmt.training - Epoch 2, Step: 4500, Batch Loss: 2.894455, Batch Acc: 0.452374, Tokens per Sec: 16361, Lr: 0.000090 2024-01-16 10:57:55,692 - INFO - joeynmt.training - Epoch 2, Step: 4600, Batch Loss: 2.756325, Batch Acc: 0.460019, Tokens per Sec: 16352, Lr: 0.000092 2024-01-16 10:59:13,508 - INFO - joeynmt.training - Epoch 2, Step: 4700, Batch Loss: 2.699669, Batch Acc: 0.467808, Tokens per Sec: 16110, Lr: 0.000094 2024-01-16 11:00:30,530 - INFO - joeynmt.training - Epoch 2, Step: 4800, Batch Loss: 2.758715, Batch Acc: 0.472943, Tokens per Sec: 16225, Lr: 0.000096 2024-01-16 11:01:46,705 - INFO - joeynmt.training - Epoch 2, Step: 4900, Batch Loss: 2.628953, Batch Acc: 0.481060, Tokens per Sec: 16395, Lr: 0.000098 2024-01-16 11:03:02,235 - INFO - joeynmt.training - Epoch 2, Step: 5000, Batch Loss: 2.529633, Batch Acc: 0.485405, Tokens per Sec: 16508, Lr: 0.000100 2024-01-16 11:03:02,235 - INFO - joeynmt.datasets - Sample random subset from dev data: n=500, seed=5042 2024-01-16 11:03:02,236 - INFO - joeynmt.prediction - Predicting 500 example(s)... (Greedy decoding with min_output_length=1, max_output_length=512, return_prob='none', generate_unk=True, repetition_penalty=-1, no_repeat_ngram_size=-1) 2024-01-16 11:04:52,807 - INFO - joeynmt.prediction - Generation took 110.5625[sec]. 2024-01-16 11:04:53,543 - INFO - joeynmt.metrics - nrefs:1|case:lc|eff:no|tok:13a|smooth:exp|version:2.4.0 2024-01-16 11:04:53,543 - INFO - joeynmt.prediction - Evaluation result (greedy): bleu: 12.10, loss: 3.10, ppl: 22.31, acc: 0.42, 0.0720[sec] 2024-01-16 11:04:53,544 - INFO - joeynmt.training - Hooray! New best validation result [bleu]! 2024-01-16 11:04:56,446 - INFO - joeynmt.training - Checkpoint saved in models/iwslt14_prompt/5000.ckpt. 2024-01-16 11:04:56,449 - INFO - joeynmt.training - Example #0 2024-01-16 11:04:56,449 - INFO - joeynmt.training - Source: You know, one of the intense pleasures of travel and one of the delights of ethnographic research is the opportunity to live amongst those who have not forgotten the old ways, who still feel their past in the wind, touch it in stones polished by rain, taste it in the bitter leaves of plants. 2024-01-16 11:04:56,449 - INFO - joeynmt.training - Reference: Vous savez, un des plaisirs intenses du voyage et un des délices de la recherche ethnographique est la possibilité de vivre parmi ceux qui n'ont pas oublié les anciennes coutumes, qui ressentent encore leur passé souffler dans le vent, qui le touchent dans les pierres polies par la pluie, le dégustent dans les feuilles amères des plantes. 2024-01-16 11:04:56,450 - INFO - joeynmt.training - Hypothesis: Vous savez, une des plus horribles de travail et une des recherches plus intenses ethniques de la recherche ethnique est la possibilité de vivre entre les plus de ceux qui ont pas entendu les plus vieux façons, qui ont toujours toujours pu le passé dans le vent, le sentir dans les vêtements, le le le le décédé, le fait dans les plantes. 2024-01-16 11:04:56,451 - DEBUG - joeynmt.training - Tokenized source: ['▁You', '▁know', ',', '▁one', '▁of', '▁the', '▁intense', '▁pleasure', 's', '▁of', '▁travel', '▁and', '▁one', '▁of', '▁the', '▁de', 'light', 's', '▁of', '▁et', 'h', 'n', 'ographic', '▁research', '▁is', '▁the', '▁opportunity', '▁to', '▁live', '▁among', 'st', '▁those', '▁who', '▁have', '▁not', '▁forgotten', '▁the', '▁old', '▁ways', ',', '▁who', '▁still', '▁feel', '▁their', '▁past', '▁in', '▁the', '▁wind', ',', '▁touch', '▁it', '▁in', '▁stones', '▁polish', 'ed', '▁by', '▁rain', ',', '▁taste', '▁it', '▁in', '▁the', '▁bitter', '▁leaves', '▁of', '▁plants', '.'] 2024-01-16 11:04:56,451 - DEBUG - joeynmt.training - Tokenized reference: ['▁Vous', '▁savez', ',', '▁un', '▁des', '▁plaisir', 's', '▁intense', 's', '▁du', '▁voyage', '▁et', '▁un', '▁des', '▁dé', 'lic', 'es', '▁de', '▁la', '▁recherche', '▁et', 'h', 'no', 'graph', 'ique', '▁est', '▁la', '▁possibilité', '▁de', '▁vivre', '▁parmi', '▁ceux', '▁qui', '▁n', "'", 'ont', '▁pas', '▁oublié', '▁les', '▁anciennes', '▁coutume', 's', ',', '▁qui', '▁ressentent', '▁encore', '▁leur', '▁passé', '▁souffle', 'r', '▁dans', '▁le', '▁vent', ',', '▁qui', '▁le', '▁touchent', '▁dans', '▁les', '▁pierres', '▁poli', 'es', '▁par', '▁la', '▁pluie', ',', '▁le', '▁dé', 'gu', 'sten', 't', '▁dans', '▁les', '▁feuilles', '▁a', 'mère', 's', '▁des', '▁plantes', '.'] 2024-01-16 11:04:56,451 - DEBUG - joeynmt.training - Tokenized hypothesis: ['', '', '▁Vous', '▁savez', ',', '▁une', '▁des', '▁plus', '▁horrible', 's', '▁de', '▁travail', '▁et', '▁une', '▁des', '▁recherches', '▁plus', '▁intense', 's', '▁et', 'h', 'n', 'ique', 's', '▁de', '▁la', '▁recherche', '▁et', 'h', 'n', 'ique', '▁est', '▁la', '▁possibilité', '▁de', '▁vivre', '▁entre', '▁les', '▁plus', '▁de', '▁ceux', '▁qui', '▁ont', '▁pas', '▁entendu', '▁les', '▁plus', '▁vieux', '▁façons', ',', '▁qui', '▁ont', '▁toujours', '▁toujours', '▁pu', '▁le', '▁passé', '▁dans', '▁le', '▁vent', ',', '▁le', '▁sentir', '▁dans', '▁les', '▁vêtements', ',', '▁le', '▁le', '▁le', '▁le', '▁dé', 'c', 'é', 'd', 'é', ',', '▁le', '▁fait', '▁dans', '▁les', '▁plantes', '.'] 2024-01-16 11:04:56,451 - INFO - joeynmt.training - Example #1 2024-01-16 11:04:56,451 - INFO - joeynmt.training - Source: Et ceci est une idée, si on y réfléchit, qui ne peut que vous remplir d'espoir. 2024-01-16 11:04:56,451 - INFO - joeynmt.training - Reference: And this is an idea, if you think about it, can only fill you with hope. 2024-01-16 11:04:56,452 - INFO - joeynmt.training - Hypothesis: And this is a idea, if we think about it, that can't be that you're going to get hope. 2024-01-16 11:04:56,452 - DEBUG - joeynmt.training - Tokenized source: ['▁Et', '▁ceci', '▁est', '▁une', '▁idée', ',', '▁si', '▁on', '▁y', '▁réfléchi', 't', ',', '▁qui', '▁ne', '▁peut', '▁que', '▁vous', '▁remplir', '▁d', "'", 'espoir', '.'] 2024-01-16 11:04:56,452 - DEBUG - joeynmt.training - Tokenized reference: ['▁And', '▁this', '▁is', '▁an', '▁idea', ',', '▁if', '▁you', '▁think', '▁about', '▁it', ',', '▁can', '▁only', '▁fill', '▁you', '▁with', '▁hope', '.'] 2024-01-16 11:04:56,453 - DEBUG - joeynmt.training - Tokenized hypothesis: ['', '▁All', '▁of', '▁these', '▁people', 's', '▁teach', '▁us', '▁that', '▁there', '▁are', '▁other', '▁ways', '▁of', '▁being', ',', '▁other', '▁ways', '▁of', '▁thinking', ',', '▁other', '▁ways', '▁of', '▁orient', 'ing', '▁yourself', '▁in', '▁the', '▁Earth', '.', '', '▁And', '▁this', '▁is', '▁a', '▁idea', ',', '▁if', '▁we', '▁think', '▁about', '▁it', ',', '▁that', '▁can', "'", 't', '▁be', '▁that', '▁you', "'", 're', '▁going', '▁to', '▁get', '▁hope', '.'] 2024-01-16 11:04:56,453 - INFO - joeynmt.training - Example #2 2024-01-16 11:04:56,453 - INFO - joeynmt.training - Source: Aujourd'hui, les innombrables cultures dans le monde constituent un tissu de vie spirituelle et culturelle qui enveloppe la planète, et qui est aussi important pour le bien-être de la planète que l'est également le tissu biologique de la vie que vous connaissez en tant que biosphère. 2024-01-16 11:04:56,453 - INFO - joeynmt.training - Reference: Now, together the myriad cultures of the world make up a web of spiritual life and cultural life that envelops the planet, and is as important to the well-being of the planet as indeed is the biological web of life that you know as a biosphere. 2024-01-16 11:04:56,453 - INFO - joeynmt.training - Hypothesis: Today, the local cultures in the world are a mental and social activity that is a planet, and that's also important for the planet, and that's also important to be the planet that the physical tissue of life that you know as the entire planet. 2024-01-16 11:04:56,454 - DEBUG - joeynmt.training - Tokenized source: ['▁A', 'ujourd', "'", 'hui', ',', '▁les', '▁', 'innombrables', '▁cultures', '▁dans', '▁le', '▁monde', '▁constituent', '▁un', '▁tissu', '▁de', '▁vie', '▁spirituelle', '▁et', '▁culturelle', '▁qui', '▁enveloppe', '▁la', '▁planète', ',', '▁et', '▁qui', '▁est', '▁aussi', '▁important', '▁pour', '▁le', '▁bien', '-', 'être', '▁de', '▁la', '▁planète', '▁que', '▁l', "'", 'est', '▁également', '▁le', '▁tissu', '▁biologique', '▁de', '▁la', '▁vie', '▁que', '▁vous', '▁connaissez', '▁en', '▁tant', '▁que', '▁biosphère', '.'] 2024-01-16 11:04:56,454 - DEBUG - joeynmt.training - Tokenized reference: ['▁Now', ',', '▁together', '▁the', '▁myriad', '▁cultures', '▁of', '▁the', '▁world', '▁make', '▁up', '▁a', '▁web', '▁of', '▁spiritual', '▁life', '▁and', '▁cultural', '▁life', '▁that', '▁en', 've', 'lop', 's', '▁the', '▁planet', ',', '▁and', '▁is', '▁as', '▁important', '▁to', '▁the', '▁well', '-', 'being', '▁of', '▁the', '▁planet', '▁as', '▁indeed', '▁is', '▁the', '▁biological', '▁web', '▁of', '▁life', '▁that', '▁you', '▁know', '▁as', '▁a', '▁biosphere', '.'] 2024-01-16 11:04:56,454 - DEBUG - joeynmt.training - Tokenized hypothesis: ['', '▁And', '▁this', '▁is', '▁an', '▁idea', ',', '▁if', '▁you', '▁think', '▁about', '▁it', ',', '▁can', '▁only', '▁fill', '▁you', '▁with', '▁hope', '.', '', '▁Today', ',', '▁the', '▁local', '▁cultures', '▁in', '▁the', '▁world', '▁are', '▁a', '▁mental', '▁and', '▁social', '▁activity', '▁that', '▁is', '▁a', '▁planet', ',', '▁and', '▁that', "'", 's', '▁also', '▁important', '▁for', '▁the', '▁planet', ',', '▁and', '▁that', "'", 's', '▁also', '▁important', '▁to', '▁be', '▁the', '▁planet', '▁that', '▁the', '▁physical', '▁tissue', '▁of', '▁life', '▁that', '▁you', '▁know', '▁as', '▁the', '▁entire', '▁planet', '.'] 2024-01-16 11:04:56,454 - INFO - joeynmt.training - Example #3 2024-01-16 11:04:56,455 - INFO - joeynmt.training - Source: C'est le symbole de tout ce que nous sommes et tout ce que nous pouvons être en tant qu'espèce dotée d'une curiosité stupéfiante. 2024-01-16 11:04:56,455 - INFO - joeynmt.training - Reference: It's the symbol of all that we are and all that we can be as an astonishingly inquisitive species. 2024-01-16 11:04:56,455 - INFO - joeynmt.training - Hypothesis: It's the symbol of all that we are and everything we can be as a remarkable self-a-a-a-old curiosity. 2024-01-16 11:04:56,456 - DEBUG - joeynmt.training - Tokenized source: ['▁C', "'", 'est', '▁le', '▁symbole', '▁de', '▁tout', '▁ce', '▁que', '▁nous', '▁sommes', '▁et', '▁tout', '▁ce', '▁que', '▁nous', '▁pouvons', '▁être', '▁en', '▁tant', '▁qu', "'", 'espèce', '▁do', 't', 'ée', '▁d', "'", 'une', '▁curiosité', '▁stupéfiant', 'e', '.'] 2024-01-16 11:04:56,456 - DEBUG - joeynmt.training - Tokenized reference: ['▁It', "'", 's', '▁the', '▁symbol', '▁of', '▁all', '▁that', '▁we', '▁are', '▁and', '▁all', '▁that', '▁we', '▁can', '▁be', '▁as', '▁an', '▁astonishing', 'ly', '▁in', 'qui', 's', 'it', 'ive', '▁species', '.'] 2024-01-16 11:04:56,456 - DEBUG - joeynmt.training - Tokenized hypothesis: ['', '▁The', '▁et', 'h', 'n', 'osphere', '▁is', '▁humanity', "'", 's', '▁great', '▁legacy', '.', '', '▁It', "'", 's', '▁the', '▁symbol', '▁of', '▁all', '▁that', '▁we', '▁are', '▁and', '▁everything', '▁we', '▁can', '▁be', '▁as', '▁a', '▁remarkable', '▁self', '-', 'a', '-', 'a', '-', 'a', '-', 'old', '▁curiosity', '.'] 2024-01-16 11:06:13,336 - INFO - joeynmt.training - Epoch 2, Step: 5100, Batch Loss: 2.479737, Batch Acc: 0.492673, Tokens per Sec: 16296, Lr: 0.000102 2024-01-16 11:07:29,015 - INFO - joeynmt.training - Epoch 2, Step: 5200, Batch Loss: 2.569900, Batch Acc: 0.498790, Tokens per Sec: 16441, Lr: 0.000104 2024-01-16 11:08:44,633 - INFO - joeynmt.training - Epoch 2, Step: 5300, Batch Loss: 2.615234, Batch Acc: 0.503549, Tokens per Sec: 16544, Lr: 0.000106 2024-01-16 11:09:59,849 - INFO - joeynmt.training - Epoch 2, Step: 5400, Batch Loss: 2.545377, Batch Acc: 0.510324, Tokens per Sec: 16550, Lr: 0.000108 2024-01-16 11:11:04,811 - INFO - joeynmt.training - Epoch 2, total training loss: 8490.68, num. of seqs: 702202, num. of tokens: 34266040, 2096.5820[sec] 2024-01-16 11:11:04,821 - INFO - joeynmt.training - EPOCH 3 2024-01-16 11:11:15,241 - INFO - joeynmt.training - Epoch 3, Step: 5500, Batch Loss: 2.378986, Batch Acc: 0.520316, Tokens per Sec: 16553, Lr: 0.000110 2024-01-16 11:12:31,730 - INFO - joeynmt.training - Epoch 3, Step: 5600, Batch Loss: 2.410851, Batch Acc: 0.524228, Tokens per Sec: 16231, Lr: 0.000112 2024-01-16 11:13:48,821 - INFO - joeynmt.training - Epoch 3, Step: 5700, Batch Loss: 2.272843, Batch Acc: 0.526903, Tokens per Sec: 16207, Lr: 0.000114 2024-01-16 11:15:05,671 - INFO - joeynmt.training - Epoch 3, Step: 5800, Batch Loss: 2.353787, Batch Acc: 0.530965, Tokens per Sec: 16212, Lr: 0.000116 2024-01-16 11:16:21,410 - INFO - joeynmt.training - Epoch 3, Step: 5900, Batch Loss: 2.347085, Batch Acc: 0.536349, Tokens per Sec: 16501, Lr: 0.000118 2024-01-16 11:17:37,733 - INFO - joeynmt.training - Epoch 3, Step: 6000, Batch Loss: 2.236222, Batch Acc: 0.539086, Tokens per Sec: 16356, Lr: 0.000120 2024-01-16 11:17:37,734 - INFO - joeynmt.datasets - Sample random subset from dev data: n=500, seed=6042 2024-01-16 11:17:37,734 - INFO - joeynmt.prediction - Predicting 500 example(s)... (Greedy decoding with min_output_length=1, max_output_length=512, return_prob='none', generate_unk=True, repetition_penalty=-1, no_repeat_ngram_size=-1) 2024-01-16 11:19:05,799 - INFO - joeynmt.prediction - Generation took 88.0561[sec]. 2024-01-16 11:19:05,938 - INFO - joeynmt.metrics - nrefs:1|case:lc|eff:no|tok:13a|smooth:exp|version:2.4.0 2024-01-16 11:19:05,938 - INFO - joeynmt.prediction - Evaluation result (greedy): bleu: 15.56, loss: 2.75, ppl: 15.61, acc: 0.47, 0.1173[sec] 2024-01-16 11:19:05,939 - INFO - joeynmt.training - Hooray! New best validation result [bleu]! 2024-01-16 11:19:08,816 - INFO - joeynmt.training - Checkpoint saved in models/iwslt14_prompt/6000.ckpt. 2024-01-16 11:19:09,260 - INFO - joeynmt.training - Example #0 2024-01-16 11:19:09,260 - INFO - joeynmt.training - Source: Nous devons faire face à la séparation inexorable de la mort, cela ne devrait donc pas nous surprendre que nous chantions, nous dansions, nous sommes tous des artistes. 2024-01-16 11:19:09,260 - INFO - joeynmt.training - Reference: We have to deal with the inexorable separation of death, so it shouldn't surprise us that we all sing, we all dance, we all have art. 2024-01-16 11:19:09,260 - INFO - joeynmt.training - Hypothesis: We need to face the untrut of death, so we shouldn't be surprised that we sing, we are all artists. 2024-01-16 11:19:09,261 - DEBUG - joeynmt.training - Tokenized source: ['▁Nous', '▁devons', '▁faire', '▁face', '▁à', '▁la', '▁séparation', '▁inexorable', '▁de', '▁la', '▁mort', ',', '▁cela', '▁ne', '▁devrait', '▁donc', '▁pas', '▁nous', '▁surprend', 're', '▁que', '▁nous', '▁chant', 'ions', ',', '▁nous', '▁dans', 'ions', ',', '▁nous', '▁sommes', '▁tous', '▁des', '▁artistes', '.'] 2024-01-16 11:19:09,261 - DEBUG - joeynmt.training - Tokenized reference: ['▁We', '▁have', '▁to', '▁deal', '▁with', '▁the', '▁inexorable', '▁separation', '▁of', '▁death', ',', '▁so', '▁it', '▁should', 'n', "'", 't', '▁surprise', '▁us', '▁that', '▁we', '▁all', '▁sing', ',', '▁we', '▁all', '▁dance', ',', '▁we', '▁all', '▁have', '▁art', '.'] 2024-01-16 11:19:09,261 - DEBUG - joeynmt.training - Tokenized hypothesis: ['', '▁We', '▁go', '▁through', '▁initiat', 'ion', '▁r', 'ites', '.', '', '▁We', '▁need', '▁to', '▁face', '▁the', '▁un', 't', 'ru', 't', '▁of', '▁death', ',', '▁so', '▁we', '▁should', 'n', "'", 't', '▁be', '▁surprised', '▁that', '▁we', '▁sing', ',', '▁we', '▁are', '▁all', '▁artists', '.'] 2024-01-16 11:19:09,261 - INFO - joeynmt.training - Example #1 2024-01-16 11:19:09,262 - INFO - joeynmt.training - Source: All of these peoples teach us that there are other ways of being, other ways of thinking, other ways of orienting yourself in the Earth. 2024-01-16 11:19:09,262 - INFO - joeynmt.training - Reference: Tous ces peuples nous enseignent qu'il y a d'autres façons d'être, d'autres façons de penser, d'autres manières de s'orienter sur Terre. 2024-01-16 11:19:09,262 - INFO - joeynmt.training - Hypothesis: Toutes ces gens nous apprennent que les autres manières d'autres pensent, d'autres façons de se connecter sur Terre. 2024-01-16 11:19:09,263 - DEBUG - joeynmt.training - Tokenized source: ['▁All', '▁of', '▁these', '▁people', 's', '▁teach', '▁us', '▁that', '▁there', '▁are', '▁other', '▁ways', '▁of', '▁being', ',', '▁other', '▁ways', '▁of', '▁thinking', ',', '▁other', '▁ways', '▁of', '▁orient', 'ing', '▁yourself', '▁in', '▁the', '▁Earth', '.'] 2024-01-16 11:19:09,263 - DEBUG - joeynmt.training - Tokenized reference: ['▁Tous', '▁ces', '▁peuple', 's', '▁nous', '▁enseigne', 'nt', '▁qu', "'", 'il', '▁y', '▁a', '▁d', "'", 'autres', '▁façons', '▁d', "'", 'être', ',', '▁d', "'", 'autres', '▁façons', '▁de', '▁penser', ',', '▁d', "'", 'autres', '▁manières', '▁de', '▁s', "'", 'orient', 'er', '▁sur', '▁Terre', '.'] 2024-01-16 11:19:09,263 - DEBUG - joeynmt.training - Tokenized hypothesis: ['', '▁Et', '▁que', '▁ce', '▁soit', '▁le', '▁Pen', 'an', '▁dans', '▁les', '▁forêts', '▁du', '▁Bo', 'r', 'né', 'o', ',', '▁ou', '▁les', '▁a', 'co', 'ly', 'tes', '▁V', 'oodoo', '▁à', '▁Haïti', ',', '▁ou', '▁bien', '▁les', '▁guerrier', 's', '▁dans', '▁le', '▁désert', '▁du', '▁K', 'ais', 'ut', '▁au', '▁nord', '▁du', '▁Kenya', ',', '▁le', '▁C', 'ur', 'ende', 'ro', '▁dans', '▁les', '▁montagnes', '▁des', '▁And', 'es', ',', '▁ou', '▁bien', '▁un', '▁car', 'ava', 'n', 's', 'é', 'ra', 'il', '▁en', '▁plein', '▁milieu', '▁du', '▁Sahara', '.', '▁A', '▁propos', ',', '▁c', "'", 'est', '▁la', '▁personne', '▁avec', '▁qui', '▁j', "'", 'ai', '▁voyagé', '▁dans', '▁le', '▁désert', '▁il', '▁y', '▁un', '▁mois', ',', '▁ou', '▁effectivement', ',', '▁le', '▁gardien', '▁de', '▁troupeau', '▁de', '▁Yak', 's', '▁sur', '▁les', '▁fla', 'n', 'c', 's', '▁du', '▁Q', 'o', 'mo', 'lang', 'ma', ',', '▁l', "'", 'Everest', ',', '▁la', '▁dé', 'esse', '▁du', '▁monde', '.', '', '▁Toutes', '▁ces', '▁gens', '▁nous', '▁apprennent', '▁que', '▁les', '▁autres', '▁manières', '▁d', "'", 'autres', '▁pensent', ',', '▁d', "'", 'autres', '▁façons', '▁de', '▁se', '▁connecter', '▁sur', '▁Terre', '.'] 2024-01-16 11:19:09,263 - INFO - joeynmt.training - Example #2 2024-01-16 11:19:09,263 - INFO - joeynmt.training - Source: Tous ces peuples nous enseignent qu'il y a d'autres façons d'être, d'autres façons de penser, d'autres manières de s'orienter sur Terre. 2024-01-16 11:19:09,263 - INFO - joeynmt.training - Reference: All of these peoples teach us that there are other ways of being, other ways of thinking, other ways of orienting yourself in the Earth. 2024-01-16 11:19:09,263 - INFO - joeynmt.training - Hypothesis: All of these people are teaching us that there are other ways of thinking, other ways of thinking, other ways of thinking, other ways of thinking about Earth. 2024-01-16 11:19:09,264 - DEBUG - joeynmt.training - Tokenized source: ['▁Tous', '▁ces', '▁peuple', 's', '▁nous', '▁enseigne', 'nt', '▁qu', "'", 'il', '▁y', '▁a', '▁d', "'", 'autres', '▁façons', '▁d', "'", 'être', ',', '▁d', "'", 'autres', '▁façons', '▁de', '▁penser', ',', '▁d', "'", 'autres', '▁manières', '▁de', '▁s', "'", 'orient', 'er', '▁sur', '▁Terre', '.'] 2024-01-16 11:19:09,264 - DEBUG - joeynmt.training - Tokenized reference: ['▁All', '▁of', '▁these', '▁people', 's', '▁teach', '▁us', '▁that', '▁there', '▁are', '▁other', '▁ways', '▁of', '▁being', ',', '▁other', '▁ways', '▁of', '▁thinking', ',', '▁other', '▁ways', '▁of', '▁orient', 'ing', '▁yourself', '▁in', '▁the', '▁Earth', '.'] 2024-01-16 11:19:09,264 - DEBUG - joeynmt.training - Tokenized hypothesis: ['', '▁And', '▁whether', '▁it', '▁is', '▁the', '▁Pen', 'an', '▁in', '▁the', '▁forests', '▁of', '▁Bo', 'r', 'ne', 'o', ',', '▁or', '▁the', '▁V', 'oodoo', '▁a', 'co', 'ly', 'tes', '▁in', '▁Haiti', ',', '▁or', '▁the', '▁warrior', 's', '▁in', '▁the', '▁K', 'ais', 'ut', '▁desert', '▁of', '▁Northern', '▁Kenya', ',', '▁the', '▁C', 'ura', 'nder', 'o', '▁in', '▁the', '▁mountains', '▁of', '▁the', '▁And', 'es', ',', '▁or', '▁a', '▁car', 'ava', 'n', 's', 'er', 'ai', '▁in', '▁the', '▁middle', '▁of', '▁the', '▁Sahara', '▁--', '▁this', '▁is', '▁incidentally', '▁the', '▁fellow', '▁that', '▁I', '▁traveled', '▁into', '▁the', '▁desert', '▁with', '▁a', '▁month', '▁ago', '▁--', '▁or', '▁indeed', '▁a', '▁y', 'ak', '▁her', 'der', '▁in', '▁the', '▁slope', 's', '▁of', '▁Q', 'o', 'mo', 'lang', 'ma', ',', '▁Everest', ',', '▁the', '▁goddess', '▁mother', '▁of', '▁the', '▁world', '.', '', '▁All', '▁of', '▁these', '▁people', '▁are', '▁teaching', '▁us', '▁that', '▁there', '▁are', '▁other', '▁ways', '▁of', '▁thinking', ',', '▁other', '▁ways', '▁of', '▁thinking', ',', '▁other', '▁ways', '▁of', '▁thinking', ',', '▁other', '▁ways', '▁of', '▁thinking', '▁about', '▁Earth', '.'] 2024-01-16 11:19:09,264 - INFO - joeynmt.training - Example #3 2024-01-16 11:19:09,265 - INFO - joeynmt.training - Source: The ethnosphere is humanity's great legacy. 2024-01-16 11:19:09,265 - INFO - joeynmt.training - Reference: L'ethnosphère est l'héritage de l'humanité. 2024-01-16 11:19:09,265 - INFO - joeynmt.training - Hypothesis: L'oshnique est une grande héritage de l'humanité. 2024-01-16 11:19:09,265 - DEBUG - joeynmt.training - Tokenized source: ['▁The', '▁et', 'h', 'n', 'osphere', '▁is', '▁humanity', "'", 's', '▁great', '▁legacy', '.'] 2024-01-16 11:19:09,266 - DEBUG - joeynmt.training - Tokenized reference: ['▁L', "'", 'e', 'th', 'no', 's', 'ph', 'ère', '▁est', '▁l', "'", 'héritage', '▁de', '▁l', "'", 'humanité', '.'] 2024-01-16 11:19:09,266 - DEBUG - joeynmt.training - Tokenized hypothesis: ['', '▁Et', '▁vous', '▁pourriez', '▁considérer', '▁ce', '▁tissu', '▁culturel', '▁de', '▁la', '▁vie', '▁en', '▁tant', '▁qu', "'", 'e', 'th', 'no', 's', 'ph', 'ère', '▁et', '▁vous', '▁pourriez', '▁définir', '▁l', "'", 'e', 'th', 'no', 's', 'ph', 'ère', '▁comme', '▁étant', '▁la', '▁somme', '▁globale', '▁de', '▁toutes', '▁les', '▁pensées', ',', '▁les', '▁rêves', ',', '▁les', '▁mythe', 's', ',', '▁les', '▁idées', ',', '▁les', '▁inspiration', 's', ',', '▁les', '▁intuition', 's', '▁engendré', 'es', '▁par', '▁l', "'", 'imagination', '▁humaine', '▁depuis', '▁l', "'", 'aube', '▁de', '▁la', '▁conscience', '.', '', '▁L', "'", 'os', 'h', 'n', 'ique', '▁est', '▁une', '▁grande', '▁héritage', '▁de', '▁l', "'", 'humanité', '.'] 2024-01-16 11:20:26,609 - INFO - joeynmt.training - Epoch 3, Step: 6100, Batch Loss: 2.241146, Batch Acc: 0.544389, Tokens per Sec: 16146, Lr: 0.000122 2024-01-16 11:21:43,934 - INFO - joeynmt.training - Epoch 3, Step: 6200, Batch Loss: 2.299156, Batch Acc: 0.545855, Tokens per Sec: 16163, Lr: 0.000124 2024-01-16 11:23:01,953 - INFO - joeynmt.training - Epoch 3, Step: 6300, Batch Loss: 2.201096, Batch Acc: 0.549491, Tokens per Sec: 15951, Lr: 0.000126 2024-01-16 11:24:18,435 - INFO - joeynmt.training - Epoch 3, Step: 6400, Batch Loss: 2.153576, Batch Acc: 0.554673, Tokens per Sec: 16325, Lr: 0.000128 2024-01-16 11:25:34,649 - INFO - joeynmt.training - Epoch 3, Step: 6500, Batch Loss: 2.176646, Batch Acc: 0.557652, Tokens per Sec: 16500, Lr: 0.000130 2024-01-16 11:26:52,200 - INFO - joeynmt.training - Epoch 3, Step: 6600, Batch Loss: 2.132968, Batch Acc: 0.560860, Tokens per Sec: 16115, Lr: 0.000132 2024-01-16 11:28:09,532 - INFO - joeynmt.training - Epoch 3, Step: 6700, Batch Loss: 2.116670, Batch Acc: 0.565132, Tokens per Sec: 16175, Lr: 0.000134 2024-01-16 11:29:26,633 - INFO - joeynmt.training - Epoch 3, Step: 6800, Batch Loss: 1.959307, Batch Acc: 0.565871, Tokens per Sec: 16193, Lr: 0.000136 2024-01-16 11:30:42,621 - INFO - joeynmt.training - Epoch 3, Step: 6900, Batch Loss: 2.128579, Batch Acc: 0.568413, Tokens per Sec: 16480, Lr: 0.000138 2024-01-16 11:31:58,676 - INFO - joeynmt.training - Epoch 3, Step: 7000, Batch Loss: 2.007592, Batch Acc: 0.570826, Tokens per Sec: 16458, Lr: 0.000140 2024-01-16 11:31:58,757 - INFO - joeynmt.datasets - Sample random subset from dev data: n=500, seed=7042 2024-01-16 11:31:58,757 - INFO - joeynmt.prediction - Predicting 500 example(s)... (Greedy decoding with min_output_length=1, max_output_length=512, return_prob='none', generate_unk=True, repetition_penalty=-1, no_repeat_ngram_size=-1) 2024-01-16 11:32:30,769 - INFO - joeynmt.prediction - Generation took 32.0113[sec]. 2024-01-16 11:32:30,999 - INFO - joeynmt.metrics - nrefs:1|case:lc|eff:no|tok:13a|smooth:exp|version:2.4.0 2024-01-16 11:32:30,999 - INFO - joeynmt.prediction - Evaluation result (greedy): bleu: 19.95, loss: 2.59, ppl: 13.38, acc: 0.50, 0.0736[sec] 2024-01-16 11:32:31,000 - INFO - joeynmt.training - Hooray! New best validation result [bleu]! 2024-01-16 11:32:33,930 - INFO - joeynmt.training - Checkpoint saved in models/iwslt14_prompt/7000.ckpt. 2024-01-16 11:32:34,379 - INFO - joeynmt.training - Example #0 2024-01-16 11:32:34,380 - INFO - joeynmt.training - Source: Wade Davis sur les cultures en voie de disparition 2024-01-16 11:32:34,380 - INFO - joeynmt.training - Reference: Wade Davis: Dreams from endangered cultures 2024-01-16 11:32:34,381 - INFO - joeynmt.training - Hypothesis: Wade Davis on the crops in the way of the land. 2024-01-16 11:32:34,381 - DEBUG - joeynmt.training - Tokenized source: ['▁Wa', 'de', '▁Davis', '▁sur', '▁les', '▁cultures', '▁en', '▁voie', '▁de', '▁disparition'] 2024-01-16 11:32:34,381 - DEBUG - joeynmt.training - Tokenized reference: ['▁Wa', 'de', '▁Davis', ':', '▁Dream', 's', '▁from', '▁endangered', '▁cultures'] 2024-01-16 11:32:34,382 - DEBUG - joeynmt.training - Tokenized hypothesis: ['', '', '▁Wa', 'de', '▁Davis', '▁on', '▁the', '▁crops', '▁in', '▁the', '▁way', '▁of', '▁the', '▁land', '.'] 2024-01-16 11:32:34,382 - INFO - joeynmt.training - Example #1 2024-01-16 11:32:34,382 - INFO - joeynmt.training - Source: Bien sûr, nous partageons tous les mêmes impératifs d'adaptation. 2024-01-16 11:32:34,382 - INFO - joeynmt.training - Reference: And of course, we all share the same adaptive imperatives. 2024-01-16 11:32:34,382 - INFO - joeynmt.training - Hypothesis: Of course, we all share the same inability. 2024-01-16 11:32:34,383 - DEBUG - joeynmt.training - Tokenized source: ['▁Bien', '▁sûr', ',', '▁nous', '▁partageons', '▁tous', '▁les', '▁mêmes', '▁im', 'pé', 'r', 'atif', 's', '▁d', "'", 'adaptation', '.'] 2024-01-16 11:32:34,383 - DEBUG - joeynmt.training - Tokenized reference: ['▁And', '▁of', '▁course', ',', '▁we', '▁all', '▁share', '▁the', '▁same', '▁adaptive', '▁imperative', 's', '.'] 2024-01-16 11:32:34,383 - DEBUG - joeynmt.training - Tokenized hypothesis: ['', '▁Just', '▁to', '▁know', '▁that', '▁Ja', 'gu', 'ar', '▁shaman', 's', '▁still', '▁journey', '▁beyond', '▁the', '▁Milk', 'y', '▁Way', ',', '▁or', '▁the', '▁myth', 's', '▁of', '▁the', '▁Inuit', '▁el', 'der', 's', '▁still', '▁resonate', '▁with', '▁meaning', ',', '▁or', '▁that', '▁in', '▁the', '▁Himalaya', ',', '▁the', '▁Buddhist', 's', '▁still', '▁pursue', '▁the', '▁breath', '▁of', '▁the', '▁D', 'ha', 'r', 'ma', ',', '▁is', '▁to', '▁really', '▁remember', '▁the', '▁central', '▁revelation', '▁of', '▁anthropology', ',', '▁and', '▁that', '▁is', '▁the', '▁idea', '▁that', '▁the', '▁world', '▁in', '▁which', '▁we', '▁live', '▁does', '▁not', '▁exist', '▁in', '▁some', '▁absolute', '▁sense', ',', '▁but', '▁is', '▁just', '▁one', '▁model', '▁of', '▁reality', ',', '▁the', '▁consequence', '▁of', '▁one', '▁particular', '▁set', '▁of', '▁adaptive', '▁choices', '▁that', '▁our', '▁line', 'age', '▁made', ',', '▁albeit', '▁successfully', ',', '▁many', '▁generations', '▁ago', '.', '', '▁Of', '▁course', ',', '▁we', '▁all', '▁share', '▁the', '▁same', '▁in', 'ability', '.'] 2024-01-16 11:32:34,383 - INFO - joeynmt.training - Example #2 2024-01-16 11:32:34,384 - INFO - joeynmt.training - Source: Nous sommes tous nés. Nous apportons nos enfants dans ce monde. 2024-01-16 11:32:34,384 - INFO - joeynmt.training - Reference: We're all born. We all bring our children into the world. 2024-01-16 11:32:34,384 - INFO - joeynmt.training - Hypothesis: We're all born. We bring our children in this world. 2024-01-16 11:32:34,384 - DEBUG - joeynmt.training - Tokenized source: ['▁Nous', '▁sommes', '▁tous', '▁né', 's', '.', '▁Nous', '▁apport', 'ons', '▁nos', '▁enfants', '▁dans', '▁ce', '▁monde', '.'] 2024-01-16 11:32:34,385 - DEBUG - joeynmt.training - Tokenized reference: ['▁We', "'", 're', '▁all', '▁born', '.', '▁We', '▁all', '▁bring', '▁our', '▁children', '▁into', '▁the', '▁world', '.'] 2024-01-16 11:32:34,385 - DEBUG - joeynmt.training - Tokenized hypothesis: ['', '▁And', '▁of', '▁course', ',', '▁we', '▁all', '▁share', '▁the', '▁same', '▁adaptive', '▁imperative', 's', '.', '', '▁We', "'", 're', '▁all', '▁born', '.', '▁We', '▁bring', '▁our', '▁children', '▁in', '▁this', '▁world', '.'] 2024-01-16 11:32:34,385 - INFO - joeynmt.training - Example #3 2024-01-16 11:32:34,385 - INFO - joeynmt.training - Source: All of these peoples teach us that there are other ways of being, other ways of thinking, other ways of orienting yourself in the Earth. 2024-01-16 11:32:34,385 - INFO - joeynmt.training - Reference: Tous ces peuples nous enseignent qu'il y a d'autres façons d'être, d'autres façons de penser, d'autres manières de s'orienter sur Terre. 2024-01-16 11:32:34,385 - INFO - joeynmt.training - Hypothesis: Tous ces gens nous apprennent qu'il y a d'autres façons de penser, d'autres façons de se se se renent dans la Terre. 2024-01-16 11:32:34,386 - DEBUG - joeynmt.training - Tokenized source: ['▁All', '▁of', '▁these', '▁people', 's', '▁teach', '▁us', '▁that', '▁there', '▁are', '▁other', '▁ways', '▁of', '▁being', ',', '▁other', '▁ways', '▁of', '▁thinking', ',', '▁other', '▁ways', '▁of', '▁orient', 'ing', '▁yourself', '▁in', '▁the', '▁Earth', '.'] 2024-01-16 11:32:34,386 - DEBUG - joeynmt.training - Tokenized reference: ['▁Tous', '▁ces', '▁peuple', 's', '▁nous', '▁enseigne', 'nt', '▁qu', "'", 'il', '▁y', '▁a', '▁d', "'", 'autres', '▁façons', '▁d', "'", 'être', ',', '▁d', "'", 'autres', '▁façons', '▁de', '▁penser', ',', '▁d', "'", 'autres', '▁manières', '▁de', '▁s', "'", 'orient', 'er', '▁sur', '▁Terre', '.'] 2024-01-16 11:32:34,386 - DEBUG - joeynmt.training - Tokenized hypothesis: ['', '▁Et', '▁que', '▁ce', '▁soit', '▁le', '▁Pen', 'an', '▁dans', '▁les', '▁forêts', '▁du', '▁Bo', 'r', 'né', 'o', ',', '▁ou', '▁les', '▁a', 'co', 'ly', 'tes', '▁V', 'oodoo', '▁à', '▁Haïti', ',', '▁ou', '▁bien', '▁les', '▁guerrier', 's', '▁dans', '▁le', '▁désert', '▁du', '▁K', 'ais', 'ut', '▁au', '▁nord', '▁du', '▁Kenya', ',', '▁le', '▁C', 'ur', 'ende', 'ro', '▁dans', '▁les', '▁montagnes', '▁des', '▁And', 'es', ',', '▁ou', '▁bien', '▁un', '▁car', 'ava', 'n', 's', 'é', 'ra', 'il', '▁en', '▁plein', '▁milieu', '▁du', '▁Sahara', '.', '▁A', '▁propos', ',', '▁c', "'", 'est', '▁la', '▁personne', '▁avec', '▁qui', '▁j', "'", 'ai', '▁voyagé', '▁dans', '▁le', '▁désert', '▁il', '▁y', '▁un', '▁mois', ',', '▁ou', '▁effectivement', ',', '▁le', '▁gardien', '▁de', '▁troupeau', '▁de', '▁Yak', 's', '▁sur', '▁les', '▁fla', 'n', 'c', 's', '▁du', '▁Q', 'o', 'mo', 'lang', 'ma', ',', '▁l', "'", 'Everest', ',', '▁la', '▁dé', 'esse', '▁du', '▁monde', '.', '', '▁Tous', '▁ces', '▁gens', '▁nous', '▁apprennent', '▁qu', "'", 'il', '▁y', '▁a', '▁d', "'", 'autres', '▁façons', '▁de', '▁penser', ',', '▁d', "'", 'autres', '▁façons', '▁de', '▁se', '▁se', '▁se', '▁re', 'n', 'ent', '▁dans', '▁la', '▁Terre', '.'] 2024-01-16 11:33:50,800 - INFO - joeynmt.training - Epoch 3, Step: 7100, Batch Loss: 2.018718, Batch Acc: 0.573801, Tokens per Sec: 16352, Lr: 0.000142 2024-01-16 11:35:08,167 - INFO - joeynmt.training - Epoch 3, Step: 7200, Batch Loss: 2.008673, Batch Acc: 0.577906, Tokens per Sec: 16143, Lr: 0.000144 2024-01-16 11:36:24,014 - INFO - joeynmt.training - Epoch 3, Step: 7300, Batch Loss: 2.023591, Batch Acc: 0.579515, Tokens per Sec: 16376, Lr: 0.000146 2024-01-16 11:37:40,631 - INFO - joeynmt.training - Epoch 3, Step: 7400, Batch Loss: 1.979726, Batch Acc: 0.581617, Tokens per Sec: 16352, Lr: 0.000148 2024-01-16 11:38:57,307 - INFO - joeynmt.training - Epoch 3, Step: 7500, Batch Loss: 1.987896, Batch Acc: 0.583732, Tokens per Sec: 16334, Lr: 0.000150 2024-01-16 11:40:14,061 - INFO - joeynmt.training - Epoch 3, Step: 7600, Batch Loss: 1.965448, Batch Acc: 0.585945, Tokens per Sec: 16194, Lr: 0.000152 2024-01-16 11:41:30,772 - INFO - joeynmt.training - Epoch 3, Step: 7700, Batch Loss: 1.953352, Batch Acc: 0.586576, Tokens per Sec: 16349, Lr: 0.000154 2024-01-16 11:42:47,159 - INFO - joeynmt.training - Epoch 3, Step: 7800, Batch Loss: 1.887127, Batch Acc: 0.590083, Tokens per Sec: 16342, Lr: 0.000156 2024-01-16 11:44:04,132 - INFO - joeynmt.training - Epoch 3, Step: 7900, Batch Loss: 1.923090, Batch Acc: 0.591658, Tokens per Sec: 16187, Lr: 0.000158 2024-01-16 11:45:20,547 - INFO - joeynmt.training - Epoch 3, Step: 8000, Batch Loss: 1.859901, Batch Acc: 0.594003, Tokens per Sec: 16400, Lr: 0.000160 2024-01-16 11:45:20,548 - INFO - joeynmt.datasets - Sample random subset from dev data: n=500, seed=8042 2024-01-16 11:45:20,548 - INFO - joeynmt.prediction - Predicting 500 example(s)... (Greedy decoding with min_output_length=1, max_output_length=512, return_prob='none', generate_unk=True, repetition_penalty=-1, no_repeat_ngram_size=-1) 2024-01-16 11:45:47,383 - INFO - joeynmt.prediction - Generation took 26.8255[sec]. 2024-01-16 11:45:47,476 - INFO - joeynmt.metrics - nrefs:1|case:lc|eff:no|tok:13a|smooth:exp|version:2.4.0 2024-01-16 11:45:47,476 - INFO - joeynmt.prediction - Evaluation result (greedy): bleu: 20.81, loss: 2.50, ppl: 12.13, acc: 0.52, 0.0729[sec] 2024-01-16 11:45:47,477 - INFO - joeynmt.training - Hooray! New best validation result [bleu]! 2024-01-16 11:45:50,297 - INFO - joeynmt.training - Checkpoint saved in models/iwslt14_prompt/8000.ckpt. 2024-01-16 11:45:50,759 - INFO - joeynmt.training - Example #0 2024-01-16 11:45:50,760 - INFO - joeynmt.training - Source: And whether it is the Penan in the forests of Borneo, or the Voodoo acolytes in Haiti, or the warriors in the Kaisut desert of Northern Kenya, the Curandero in the mountains of the Andes, or a caravanserai in the middle of the Sahara -- this is incidentally the fellow that I traveled into the desert with a month ago -- or indeed a yak herder in the slopes of Qomolangma, Everest, the goddess mother of the world. 2024-01-16 11:45:50,760 - INFO - joeynmt.training - Reference: Et que ce soit le Penan dans les forêts du Bornéo, ou les acolytes Voodoo à Haïti, ou bien les guerriers dans le désert du Kaisut au nord du Kenya, le Curendero dans les montagnes des Andes, ou bien un caravansérail en plein milieu du Sahara. A propos, c'est la personne avec qui j'ai voyagé dans le désert il y un mois, ou effectivement, le gardien de troupeau de Yaks sur les flancs du Qomolangma, l'Everest, la déesse du monde. 2024-01-16 11:45:50,760 - INFO - joeynmt.training - Hypothesis: Et si c'est le Penan dans les forêts de Borneo, ou le Voodoo d'uncolytes en Haïti, ou les guerriers dans le désert de Kaisut du Nord du Kenya, le Curandero dans les montagnes de La Etes, ou un caravanserai au milieu du Sahara -- c'est par le moment où je suis entré dans le désert avec un mois -- ou en fait un crosssssssssssssssser dans les pentes de Qomomalma, la mère du monde. 2024-01-16 11:45:50,761 - DEBUG - joeynmt.training - Tokenized source: ['▁And', '▁whether', '▁it', '▁is', '▁the', '▁Pen', 'an', '▁in', '▁the', '▁forests', '▁of', '▁Bo', 'r', 'ne', 'o', ',', '▁or', '▁the', '▁V', 'oodoo', '▁a', 'co', 'ly', 'tes', '▁in', '▁Haiti', ',', '▁or', '▁the', '▁warrior', 's', '▁in', '▁the', '▁K', 'ais', 'ut', '▁desert', '▁of', '▁Northern', '▁Kenya', ',', '▁the', '▁C', 'ura', 'nder', 'o', '▁in', '▁the', '▁mountains', '▁of', '▁the', '▁And', 'es', ',', '▁or', '▁a', '▁car', 'ava', 'n', 's', 'er', 'ai', '▁in', '▁the', '▁middle', '▁of', '▁the', '▁Sahara', '▁--', '▁this', '▁is', '▁incidentally', '▁the', '▁fellow', '▁that', '▁I', '▁traveled', '▁into', '▁the', '▁desert', '▁with', '▁a', '▁month', '▁ago', '▁--', '▁or', '▁indeed', '▁a', '▁y', 'ak', '▁her', 'der', '▁in', '▁the', '▁slope', 's', '▁of', '▁Q', 'o', 'mo', 'lang', 'ma', ',', '▁Everest', ',', '▁the', '▁goddess', '▁mother', '▁of', '▁the', '▁world', '.'] 2024-01-16 11:45:50,762 - DEBUG - joeynmt.training - Tokenized reference: ['▁Et', '▁que', '▁ce', '▁soit', '▁le', '▁Pen', 'an', '▁dans', '▁les', '▁forêts', '▁du', '▁Bo', 'r', 'né', 'o', ',', '▁ou', '▁les', '▁a', 'co', 'ly', 'tes', '▁V', 'oodoo', '▁à', '▁Haïti', ',', '▁ou', '▁bien', '▁les', '▁guerrier', 's', '▁dans', '▁le', '▁désert', '▁du', '▁K', 'ais', 'ut', '▁au', '▁nord', '▁du', '▁Kenya', ',', '▁le', '▁C', 'ur', 'ende', 'ro', '▁dans', '▁les', '▁montagnes', '▁des', '▁And', 'es', ',', '▁ou', '▁bien', '▁un', '▁car', 'ava', 'n', 's', 'é', 'ra', 'il', '▁en', '▁plein', '▁milieu', '▁du', '▁Sahara', '.', '▁A', '▁propos', ',', '▁c', "'", 'est', '▁la', '▁personne', '▁avec', '▁qui', '▁j', "'", 'ai', '▁voyagé', '▁dans', '▁le', '▁désert', '▁il', '▁y', '▁un', '▁mois', ',', '▁ou', '▁effectivement', ',', '▁le', '▁gardien', '▁de', '▁troupeau', '▁de', '▁Yak', 's', '▁sur', '▁les', '▁fla', 'n', 'c', 's', '▁du', '▁Q', 'o', 'mo', 'lang', 'ma', ',', '▁l', "'", 'Everest', ',', '▁la', '▁dé', 'esse', '▁du', '▁monde', '.'] 2024-01-16 11:45:50,762 - DEBUG - joeynmt.training - Tokenized hypothesis: ['', '▁Par', '▁contre', ',', '▁la', '▁cadence', '▁exceptionnel', 'le', '▁de', '▁la', '▁chanson', '▁est', '▁intéressante', ',', '▁le', '▁rythme', '▁de', '▁la', '▁dans', 'e', '▁dans', '▁toutes', '▁les', '▁cultures', '.', '', '▁Et', '▁si', '▁c', "'", 'est', '▁le', '▁Pen', 'an', '▁dans', '▁les', '▁forêts', '▁de', '▁Bo', 'r', 'ne', 'o', ',', '▁ou', '▁le', '▁V', 'oodoo', '▁d', "'", 'un', 'co', 'ly', 'tes', '▁en', '▁Haïti', ',', '▁ou', '▁les', '▁guerrier', 's', '▁dans', '▁le', '▁désert', '▁de', '▁K', 'ais', 'ut', '▁du', '▁Nord', '▁du', '▁Kenya', ',', '▁le', '▁C', 'ura', 'nder', 'o', '▁dans', '▁les', '▁montagnes', '▁de', '▁La', '▁Et', 'es', ',', '▁ou', '▁un', '▁car', 'ava', 'n', 's', 'er', 'ai', '▁au', '▁milieu', '▁du', '▁Sahara', '▁--', '▁c', "'", 'est', '▁par', '▁le', '▁moment', '▁où', '▁je', '▁suis', '▁entré', '▁dans', '▁le', '▁désert', '▁avec', '▁un', '▁mois', '▁--', '▁ou', '▁en', '▁fait', '▁un', '▁c', 'ro', 's', 's', 's', 's', 's', 's', 's', 's', 's', 's', 's', 's', 's', 's', 's', 's', 'er', '▁dans', '▁les', '▁pente', 's', '▁de', '▁Q', 'o', 'mo', 'mal', 'ma', ',', '▁la', '▁mère', '▁du', '▁monde', '.'] 2024-01-16 11:45:50,762 - INFO - joeynmt.training - Example #1 2024-01-16 11:45:50,762 - INFO - joeynmt.training - Source: All of these peoples teach us that there are other ways of being, other ways of thinking, other ways of orienting yourself in the Earth. 2024-01-16 11:45:50,762 - INFO - joeynmt.training - Reference: Tous ces peuples nous enseignent qu'il y a d'autres façons d'être, d'autres façons de penser, d'autres manières de s'orienter sur Terre. 2024-01-16 11:45:50,762 - INFO - joeynmt.training - Hypothesis: Tous ces personnes nous apprennent qu'il y a d'autres façons d'être, d'autres façons de penser, d'autres façons de se transformer dans la Terre. 2024-01-16 11:45:50,763 - DEBUG - joeynmt.training - Tokenized source: ['▁All', '▁of', '▁these', '▁people', 's', '▁teach', '▁us', '▁that', '▁there', '▁are', '▁other', '▁ways', '▁of', '▁being', ',', '▁other', '▁ways', '▁of', '▁thinking', ',', '▁other', '▁ways', '▁of', '▁orient', 'ing', '▁yourself', '▁in', '▁the', '▁Earth', '.'] 2024-01-16 11:45:50,763 - DEBUG - joeynmt.training - Tokenized reference: ['▁Tous', '▁ces', '▁peuple', 's', '▁nous', '▁enseigne', 'nt', '▁qu', "'", 'il', '▁y', '▁a', '▁d', "'", 'autres', '▁façons', '▁d', "'", 'être', ',', '▁d', "'", 'autres', '▁façons', '▁de', '▁penser', ',', '▁d', "'", 'autres', '▁manières', '▁de', '▁s', "'", 'orient', 'er', '▁sur', '▁Terre', '.'] 2024-01-16 11:45:50,763 - DEBUG - joeynmt.training - Tokenized hypothesis: ['', '▁Et', '▁que', '▁ce', '▁soit', '▁le', '▁Pen', 'an', '▁dans', '▁les', '▁forêts', '▁du', '▁Bo', 'r', 'né', 'o', ',', '▁ou', '▁les', '▁a', 'co', 'ly', 'tes', '▁V', 'oodoo', '▁à', '▁Haïti', ',', '▁ou', '▁bien', '▁les', '▁guerrier', 's', '▁dans', '▁le', '▁désert', '▁du', '▁K', 'ais', 'ut', '▁au', '▁nord', '▁du', '▁Kenya', ',', '▁le', '▁C', 'ur', 'ende', 'ro', '▁dans', '▁les', '▁montagnes', '▁des', '▁And', 'es', ',', '▁ou', '▁bien', '▁un', '▁car', 'ava', 'n', 's', 'é', 'ra', 'il', '▁en', '▁plein', '▁milieu', '▁du', '▁Sahara', '.', '▁A', '▁propos', ',', '▁c', "'", 'est', '▁la', '▁personne', '▁avec', '▁qui', '▁j', "'", 'ai', '▁voyagé', '▁dans', '▁le', '▁désert', '▁il', '▁y', '▁un', '▁mois', ',', '▁ou', '▁effectivement', ',', '▁le', '▁gardien', '▁de', '▁troupeau', '▁de', '▁Yak', 's', '▁sur', '▁les', '▁fla', 'n', 'c', 's', '▁du', '▁Q', 'o', 'mo', 'lang', 'ma', ',', '▁l', "'", 'Everest', ',', '▁la', '▁dé', 'esse', '▁du', '▁monde', '.', '', '▁Tous', '▁ces', '▁personnes', '▁nous', '▁apprennent', '▁qu', "'", 'il', '▁y', '▁a', '▁d', "'", 'autres', '▁façons', '▁d', "'", 'être', ',', '▁d', "'", 'autres', '▁façons', '▁de', '▁penser', ',', '▁d', "'", 'autres', '▁façons', '▁de', '▁se', '▁transformer', '▁dans', '▁la', '▁Terre', '.'] 2024-01-16 11:45:50,763 - INFO - joeynmt.training - Example #2 2024-01-16 11:45:50,763 - INFO - joeynmt.training - Source: Aujourd'hui, les innombrables cultures dans le monde constituent un tissu de vie spirituelle et culturelle qui enveloppe la planète, et qui est aussi important pour le bien-être de la planète que l'est également le tissu biologique de la vie que vous connaissez en tant que biosphère. 2024-01-16 11:45:50,764 - INFO - joeynmt.training - Reference: Now, together the myriad cultures of the world make up a web of spiritual life and cultural life that envelops the planet, and is as important to the well-being of the planet as indeed is the biological web of life that you know as a biosphere. 2024-01-16 11:45:50,764 - INFO - joeynmt.training - Hypothesis: Today, the countless cultures in the world are a tissue of spiritual and cultural life that's also important for the well-being of the planet that's also the biological fabric of life that you're familiar with as the biosphere. 2024-01-16 11:45:50,764 - DEBUG - joeynmt.training - Tokenized source: ['▁A', 'ujourd', "'", 'hui', ',', '▁les', '▁', 'innombrables', '▁cultures', '▁dans', '▁le', '▁monde', '▁constituent', '▁un', '▁tissu', '▁de', '▁vie', '▁spirituelle', '▁et', '▁culturelle', '▁qui', '▁enveloppe', '▁la', '▁planète', ',', '▁et', '▁qui', '▁est', '▁aussi', '▁important', '▁pour', '▁le', '▁bien', '-', 'être', '▁de', '▁la', '▁planète', '▁que', '▁l', "'", 'est', '▁également', '▁le', '▁tissu', '▁biologique', '▁de', '▁la', '▁vie', '▁que', '▁vous', '▁connaissez', '▁en', '▁tant', '▁que', '▁biosphère', '.'] 2024-01-16 11:45:50,765 - DEBUG - joeynmt.training - Tokenized reference: ['▁Now', ',', '▁together', '▁the', '▁myriad', '▁cultures', '▁of', '▁the', '▁world', '▁make', '▁up', '▁a', '▁web', '▁of', '▁spiritual', '▁life', '▁and', '▁cultural', '▁life', '▁that', '▁en', 've', 'lop', 's', '▁the', '▁planet', ',', '▁and', '▁is', '▁as', '▁important', '▁to', '▁the', '▁well', '-', 'being', '▁of', '▁the', '▁planet', '▁as', '▁indeed', '▁is', '▁the', '▁biological', '▁web', '▁of', '▁life', '▁that', '▁you', '▁know', '▁as', '▁a', '▁biosphere', '.'] 2024-01-16 11:45:50,765 - DEBUG - joeynmt.training - Tokenized hypothesis: ['', '▁And', '▁this', '▁is', '▁an', '▁idea', ',', '▁if', '▁you', '▁think', '▁about', '▁it', ',', '▁can', '▁only', '▁fill', '▁you', '▁with', '▁hope', '.', '', '▁Today', ',', '▁the', '▁countless', '▁cultures', '▁in', '▁the', '▁world', '▁are', '▁a', '▁tissue', '▁of', '▁spiritual', '▁and', '▁cultural', '▁life', '▁that', "'", 's', '▁also', '▁important', '▁for', '▁the', '▁well', '-', 'being', '▁of', '▁the', '▁planet', '▁that', "'", 's', '▁also', '▁the', '▁biological', '▁fabric', '▁of', '▁life', '▁that', '▁you', "'", 're', '▁familiar', '▁with', '▁as', '▁the', '▁biosphere', '.'] 2024-01-16 11:45:50,765 - INFO - joeynmt.training - Example #3 2024-01-16 11:45:50,765 - INFO - joeynmt.training - Source: No biologists, for example, would dare suggest that 50 percent of all species or more have been or are on the brink of extinction because it simply is not true, and yet that -- the most apocalyptic scenario in the realm of biological diversity -- scarcely approaches what we know to be the most optimistic scenario in the realm of cultural diversity. 2024-01-16 11:45:50,765 - INFO - joeynmt.training - Reference: Aucun biologiste, par exemple, n'oserait suggérer que 50% ou plus de toutes les espèces ont été ou sont à deux doigts de l'extinction parce que ce n'est tout simplement pas vrai, et pourtant -- que le scénario le plus apocalyptique dans le royaume de la diversité biologique -- se rapproche rarement de ce que nous considérons comme le scénario le plus optimiste au sein de la diversité culturelle. 2024-01-16 11:45:50,765 - INFO - joeynmt.training - Hypothesis: Pas des biologistes, par exemple, suggèreraient que 50 % de toutes les espèces ou plus ont été ou moins d'extinctions parce qu'il ne s'est pas vrai, et pourtant, le scénario le plus aposénique dans le domaine de la diversité biologique -- des approches rarement des choses que nous savons être le scénario le plus optimiste dans le domaine de la diversité culturelle. 2024-01-16 11:45:50,766 - DEBUG - joeynmt.training - Tokenized source: ['▁No', '▁biologists', ',', '▁for', '▁example', ',', '▁would', '▁da', 're', '▁suggest', '▁that', '▁50', '▁percent', '▁of', '▁all', '▁species', '▁or', '▁more', '▁have', '▁been', '▁or', '▁are', '▁on', '▁the', '▁brin', 'k', '▁of', '▁extinction', '▁because', '▁it', '▁simply', '▁is', '▁not', '▁true', ',', '▁and', '▁yet', '▁that', '▁--', '▁the', '▁most', '▁apocalyptic', '▁scenario', '▁in', '▁the', '▁realm', '▁of', '▁biological', '▁diversity', '▁--', '▁scarce', 'ly', '▁approaches', '▁what', '▁we', '▁know', '▁to', '▁be', '▁the', '▁most', '▁optimistic', '▁scenario', '▁in', '▁the', '▁realm', '▁of', '▁cultural', '▁diversity', '.'] 2024-01-16 11:45:50,766 - DEBUG - joeynmt.training - Tokenized reference: ['▁Aucun', '▁biologist', 'e', ',', '▁par', '▁exemple', ',', '▁n', "'", 'ose', 'rait', '▁suggérer', '▁que', '▁50%', '▁ou', '▁plus', '▁de', '▁toutes', '▁les', '▁espèces', '▁ont', '▁été', '▁ou', '▁sont', '▁à', '▁deux', '▁doigts', '▁de', '▁l', "'", 'extinction', '▁parce', '▁que', '▁ce', '▁n', "'", 'est', '▁tout', '▁simplement', '▁pas', '▁vrai', ',', '▁et', '▁pourtant', '▁--', '▁que', '▁le', '▁scénario', '▁le', '▁plus', '▁a', 'po', 'ca', 'ly', 'p', 'tique', '▁dans', '▁le', '▁royaume', '▁de', '▁la', '▁diversité', '▁biologique', '▁--', '▁se', '▁rapproche', '▁rarement', '▁de', '▁ce', '▁que', '▁nous', '▁considér', 'ons', '▁comme', '▁le', '▁scénario', '▁le', '▁plus', '▁optimiste', '▁au', '▁sein', '▁de', '▁la', '▁diversité', '▁culturelle', '.'] 2024-01-16 11:45:50,766 - DEBUG - joeynmt.training - Tokenized hypothesis: ['', '▁Et', '▁lorsque', '▁la', '▁biosphère', '▁fut', '▁sérieusement', '▁é', 'ro', 'd', 'ée', ',', '▁l', "'", 'e', 'th', 'no', 's', 'ph', 'ère', '▁l', "'", 'a', '▁été', '▁également', '▁--', '▁et', '▁peut', '-', 'être', '▁bien', '▁plus', '▁rapidement', '.', '', '▁Pas', '▁des', '▁biologistes', ',', '▁par', '▁exemple', ',', '▁suggère', 'raient', '▁que', '▁50', '▁%', '▁de', '▁toutes', '▁les', '▁espèces', '▁ou', '▁plus', '▁ont', '▁été', '▁ou', '▁moins', '▁d', "'", 'extinction', 's', '▁parce', '▁qu', "'", 'il', '▁ne', '▁s', "'", 'est', '▁pas', '▁vrai', ',', '▁et', '▁pourtant', ',', '▁le', '▁scénario', '▁le', '▁plus', '▁a', 'po', 's', 'é', 'n', 'ique', '▁dans', '▁le', '▁domaine', '▁de', '▁la', '▁diversité', '▁biologique', '▁--', '▁des', '▁approche', 's', '▁rare', 'ment', '▁des', '▁choses', '▁que', '▁nous', '▁savons', '▁être', '▁le', '▁scénario', '▁le', '▁plus', '▁optimiste', '▁dans', '▁le', '▁domaine', '▁de', '▁la', '▁diversité', '▁culturelle', '.'] 2024-01-16 11:47:07,090 - INFO - joeynmt.training - Epoch 3, Step: 8100, Batch Loss: 1.926306, Batch Acc: 0.595788, Tokens per Sec: 16411, Lr: 0.000162 2024-01-16 11:48:24,177 - INFO - joeynmt.training - Epoch 3, Step: 8200, Batch Loss: 1.871999, Batch Acc: 0.595788, Tokens per Sec: 16213, Lr: 0.000164 2024-01-16 11:48:46,283 - INFO - joeynmt.training - Epoch 3, total training loss: 5782.68, num. of seqs: 702202, num. of tokens: 34266040, 2103.9491[sec] 2024-01-16 11:48:46,293 - INFO - joeynmt.training - EPOCH 4 2024-01-16 11:49:41,738 - INFO - joeynmt.training - Epoch 4, Step: 8300, Batch Loss: 1.842270, Batch Acc: 0.607311, Tokens per Sec: 16128, Lr: 0.000166 2024-01-16 11:50:58,105 - INFO - joeynmt.training - Epoch 4, Step: 8400, Batch Loss: 1.833386, Batch Acc: 0.607119, Tokens per Sec: 16340, Lr: 0.000168 2024-01-16 11:52:14,012 - INFO - joeynmt.training - Epoch 4, Step: 8500, Batch Loss: 1.851339, Batch Acc: 0.610090, Tokens per Sec: 16424, Lr: 0.000170 2024-01-16 11:53:32,935 - INFO - joeynmt.training - Epoch 4, Step: 8600, Batch Loss: 1.864623, Batch Acc: 0.610212, Tokens per Sec: 15913, Lr: 0.000172 2024-01-16 11:54:49,584 - INFO - joeynmt.training - Epoch 4, Step: 8700, Batch Loss: 1.802165, Batch Acc: 0.610850, Tokens per Sec: 16242, Lr: 0.000174 2024-01-16 11:56:05,670 - INFO - joeynmt.training - Epoch 4, Step: 8800, Batch Loss: 1.837649, Batch Acc: 0.610854, Tokens per Sec: 16450, Lr: 0.000176 2024-01-16 11:57:22,109 - INFO - joeynmt.training - Epoch 4, Step: 8900, Batch Loss: 1.783293, Batch Acc: 0.613373, Tokens per Sec: 16359, Lr: 0.000178 2024-01-16 11:58:38,221 - INFO - joeynmt.training - Epoch 4, Step: 9000, Batch Loss: 1.803631, Batch Acc: 0.614296, Tokens per Sec: 16425, Lr: 0.000180 2024-01-16 11:58:38,222 - INFO - joeynmt.datasets - Sample random subset from dev data: n=500, seed=9042 2024-01-16 11:58:38,222 - INFO - joeynmt.prediction - Predicting 500 example(s)... (Greedy decoding with min_output_length=1, max_output_length=512, return_prob='none', generate_unk=True, repetition_penalty=-1, no_repeat_ngram_size=-1) 2024-01-16 11:59:17,594 - INFO - joeynmt.prediction - Generation took 39.3638[sec]. 2024-01-16 11:59:17,732 - INFO - joeynmt.metrics - nrefs:1|case:lc|eff:no|tok:13a|smooth:exp|version:2.4.0 2024-01-16 11:59:17,732 - INFO - joeynmt.prediction - Evaluation result (greedy): bleu: 20.65, loss: 2.33, ppl: 10.24, acc: 0.54, 0.1195[sec] 2024-01-16 11:59:20,650 - INFO - joeynmt.training - Checkpoint saved in models/iwslt14_prompt/9000.ckpt. 2024-01-16 11:59:21,048 - INFO - joeynmt.training - Example #0 2024-01-16 11:59:21,050 - INFO - joeynmt.training - Source: Nous procédons à des rites d'initiations. 2024-01-16 11:59:21,050 - INFO - joeynmt.training - Reference: We go through initiation rites. 2024-01-16 11:59:21,050 - INFO - joeynmt.training - Hypothesis: We're doing the Israelis. 2024-01-16 11:59:21,051 - DEBUG - joeynmt.training - Tokenized source: ['▁Nous', '▁procéd', 'ons', '▁à', '▁des', '▁r', 'ites', '▁d', "'", 'ini', 'ti', 'ations', '.'] 2024-01-16 11:59:21,051 - DEBUG - joeynmt.training - Tokenized reference: ['▁We', '▁go', '▁through', '▁initiat', 'ion', '▁r', 'ites', '.'] 2024-01-16 11:59:21,051 - DEBUG - joeynmt.training - Tokenized hypothesis: ['', '▁We', "'", 're', '▁all', '▁born', '.', '▁We', '▁all', '▁bring', '▁our', '▁children', '▁into', '▁the', '▁world', '.', '', '▁We', "'", 're', '▁doing', '▁the', '▁Israeli', 's', '.'] 2024-01-16 11:59:21,051 - INFO - joeynmt.training - Example #1 2024-01-16 11:59:21,051 - INFO - joeynmt.training - Source: Now, together the myriad cultures of the world make up a web of spiritual life and cultural life that envelops the planet, and is as important to the well-being of the planet as indeed is the biological web of life that you know as a biosphere. 2024-01-16 11:59:21,051 - INFO - joeynmt.training - Reference: Aujourd'hui, les innombrables cultures dans le monde constituent un tissu de vie spirituelle et culturelle qui enveloppe la planète, et qui est aussi important pour le bien-être de la planète que l'est également le tissu biologique de la vie que vous connaissez en tant que biosphère. 2024-01-16 11:59:21,051 - INFO - joeynmt.training - Hypothesis: Maintenant, ensemble les nombreuses cultures du monde font un web de vie spirituelle et de la vie culturelle qui a envelopé la planète, et qui sont aussi importantes pour le bien-être de la planète, en fait, c'est le web biologique de la vie que vous connaissez comme une biosphère. 2024-01-16 11:59:21,052 - DEBUG - joeynmt.training - Tokenized source: ['▁Now', ',', '▁together', '▁the', '▁myriad', '▁cultures', '▁of', '▁the', '▁world', '▁make', '▁up', '▁a', '▁web', '▁of', '▁spiritual', '▁life', '▁and', '▁cultural', '▁life', '▁that', '▁en', 've', 'lop', 's', '▁the', '▁planet', ',', '▁and', '▁is', '▁as', '▁important', '▁to', '▁the', '▁well', '-', 'being', '▁of', '▁the', '▁planet', '▁as', '▁indeed', '▁is', '▁the', '▁biological', '▁web', '▁of', '▁life', '▁that', '▁you', '▁know', '▁as', '▁a', '▁biosphere', '.'] 2024-01-16 11:59:21,052 - DEBUG - joeynmt.training - Tokenized reference: ['▁A', 'ujourd', "'", 'hui', ',', '▁les', '▁', 'innombrables', '▁cultures', '▁dans', '▁le', '▁monde', '▁constituent', '▁un', '▁tissu', '▁de', '▁vie', '▁spirituelle', '▁et', '▁culturelle', '▁qui', '▁enveloppe', '▁la', '▁planète', ',', '▁et', '▁qui', '▁est', '▁aussi', '▁important', '▁pour', '▁le', '▁bien', '-', 'être', '▁de', '▁la', '▁planète', '▁que', '▁l', "'", 'est', '▁également', '▁le', '▁tissu', '▁biologique', '▁de', '▁la', '▁vie', '▁que', '▁vous', '▁connaissez', '▁en', '▁tant', '▁que', '▁biosphère', '.'] 2024-01-16 11:59:21,052 - DEBUG - joeynmt.training - Tokenized hypothesis: ['', '▁Et', '▁ceci', '▁est', '▁une', '▁idée', ',', '▁si', '▁on', '▁y', '▁réfléchi', 't', ',', '▁qui', '▁ne', '▁peut', '▁que', '▁vous', '▁remplir', '▁d', "'", 'espoir', '.', '', '▁Maintenant', ',', '▁ensemble', '▁les', '▁nombreuses', '▁cultures', '▁du', '▁monde', '▁font', '▁un', '▁web', '▁de', '▁vie', '▁spirituelle', '▁et', '▁de', '▁la', '▁vie', '▁culturelle', '▁qui', '▁a', '▁en', 've', 'lop', 'é', '▁la', '▁planète', ',', '▁et', '▁qui', '▁sont', '▁aussi', '▁importantes', '▁pour', '▁le', '▁bien', '-', 'être', '▁de', '▁la', '▁planète', ',', '▁en', '▁fait', ',', '▁c', "'", 'est', '▁le', '▁web', '▁biologique', '▁de', '▁la', '▁vie', '▁que', '▁vous', '▁connaissez', '▁comme', '▁une', '▁biosphère', '.'] 2024-01-16 11:59:21,052 - INFO - joeynmt.training - Example #2 2024-01-16 11:59:21,053 - INFO - joeynmt.training - Source: The ethnosphere is humanity's great legacy. 2024-01-16 11:59:21,053 - INFO - joeynmt.training - Reference: L'ethnosphère est l'héritage de l'humanité. 2024-01-16 11:59:21,053 - INFO - joeynmt.training - Hypothesis: L'arténique est l'héritage de l'humanité. 2024-01-16 11:59:21,054 - DEBUG - joeynmt.training - Tokenized source: ['▁The', '▁et', 'h', 'n', 'osphere', '▁is', '▁humanity', "'", 's', '▁great', '▁legacy', '.'] 2024-01-16 11:59:21,054 - DEBUG - joeynmt.training - Tokenized reference: ['▁L', "'", 'e', 'th', 'no', 's', 'ph', 'ère', '▁est', '▁l', "'", 'héritage', '▁de', '▁l', "'", 'humanité', '.'] 2024-01-16 11:59:21,054 - DEBUG - joeynmt.training - Tokenized hypothesis: ['', '▁Et', '▁vous', '▁pourriez', '▁considérer', '▁ce', '▁tissu', '▁culturel', '▁de', '▁la', '▁vie', '▁en', '▁tant', '▁qu', "'", 'e', 'th', 'no', 's', 'ph', 'ère', '▁et', '▁vous', '▁pourriez', '▁définir', '▁l', "'", 'e', 'th', 'no', 's', 'ph', 'ère', '▁comme', '▁étant', '▁la', '▁somme', '▁globale', '▁de', '▁toutes', '▁les', '▁pensées', ',', '▁les', '▁rêves', ',', '▁les', '▁mythe', 's', ',', '▁les', '▁idées', ',', '▁les', '▁inspiration', 's', ',', '▁les', '▁intuition', 's', '▁engendré', 'es', '▁par', '▁l', "'", 'imagination', '▁humaine', '▁depuis', '▁l', "'", 'aube', '▁de', '▁la', '▁conscience', '.', '', '▁L', "'", 'art', 'é', 'n', 'ique', '▁est', '▁l', "'", 'héritage', '▁de', '▁l', "'", 'humanité', '.'] 2024-01-16 11:59:21,054 - INFO - joeynmt.training - Example #3 2024-01-16 11:59:21,054 - INFO - joeynmt.training - Source: No biologists, for example, would dare suggest that 50 percent of all species or more have been or are on the brink of extinction because it simply is not true, and yet that -- the most apocalyptic scenario in the realm of biological diversity -- scarcely approaches what we know to be the most optimistic scenario in the realm of cultural diversity. 2024-01-16 11:59:21,054 - INFO - joeynmt.training - Reference: Aucun biologiste, par exemple, n'oserait suggérer que 50% ou plus de toutes les espèces ont été ou sont à deux doigts de l'extinction parce que ce n'est tout simplement pas vrai, et pourtant -- que le scénario le plus apocalyptique dans le royaume de la diversité biologique -- se rapproche rarement de ce que nous considérons comme le scénario le plus optimiste au sein de la diversité culturelle. 2024-01-16 11:59:21,054 - INFO - joeynmt.training - Hypothesis: Pas des biologistes, par exemple, suggèrent que 50 pour cent de toutes les espèces ou plus ont été ou sont sur le bord de l'extinction parce que ce n'est pas vrai, et pourtant, le scénario le plus aplypénique dans le domaine de la diversité biologique -- des approches rarement ce que nous savons être le scénario le plus optimiste dans le domaine de la diversité culturelle. 2024-01-16 11:59:21,055 - DEBUG - joeynmt.training - Tokenized source: ['▁No', '▁biologists', ',', '▁for', '▁example', ',', '▁would', '▁da', 're', '▁suggest', '▁that', '▁50', '▁percent', '▁of', '▁all', '▁species', '▁or', '▁more', '▁have', '▁been', '▁or', '▁are', '▁on', '▁the', '▁brin', 'k', '▁of', '▁extinction', '▁because', '▁it', '▁simply', '▁is', '▁not', '▁true', ',', '▁and', '▁yet', '▁that', '▁--', '▁the', '▁most', '▁apocalyptic', '▁scenario', '▁in', '▁the', '▁realm', '▁of', '▁biological', '▁diversity', '▁--', '▁scarce', 'ly', '▁approaches', '▁what', '▁we', '▁know', '▁to', '▁be', '▁the', '▁most', '▁optimistic', '▁scenario', '▁in', '▁the', '▁realm', '▁of', '▁cultural', '▁diversity', '.'] 2024-01-16 11:59:21,055 - DEBUG - joeynmt.training - Tokenized reference: ['▁Aucun', '▁biologist', 'e', ',', '▁par', '▁exemple', ',', '▁n', "'", 'ose', 'rait', '▁suggérer', '▁que', '▁50%', '▁ou', '▁plus', '▁de', '▁toutes', '▁les', '▁espèces', '▁ont', '▁été', '▁ou', '▁sont', '▁à', '▁deux', '▁doigts', '▁de', '▁l', "'", 'extinction', '▁parce', '▁que', '▁ce', '▁n', "'", 'est', '▁tout', '▁simplement', '▁pas', '▁vrai', ',', '▁et', '▁pourtant', '▁--', '▁que', '▁le', '▁scénario', '▁le', '▁plus', '▁a', 'po', 'ca', 'ly', 'p', 'tique', '▁dans', '▁le', '▁royaume', '▁de', '▁la', '▁diversité', '▁biologique', '▁--', '▁se', '▁rapproche', '▁rarement', '▁de', '▁ce', '▁que', '▁nous', '▁considér', 'ons', '▁comme', '▁le', '▁scénario', '▁le', '▁plus', '▁optimiste', '▁au', '▁sein', '▁de', '▁la', '▁diversité', '▁culturelle', '.'] 2024-01-16 11:59:21,055 - DEBUG - joeynmt.training - Tokenized hypothesis: ['', '▁Et', '▁lorsque', '▁la', '▁biosphère', '▁fut', '▁sérieusement', '▁é', 'ro', 'd', 'ée', ',', '▁l', "'", 'e', 'th', 'no', 's', 'ph', 'ère', '▁l', "'", 'a', '▁été', '▁également', '▁--', '▁et', '▁peut', '-', 'être', '▁bien', '▁plus', '▁rapidement', '.', '', '▁Pas', '▁des', '▁biologistes', ',', '▁par', '▁exemple', ',', '▁suggère', 'nt', '▁que', '▁50', '▁pour', '▁cent', '▁de', '▁toutes', '▁les', '▁espèces', '▁ou', '▁plus', '▁ont', '▁été', '▁ou', '▁sont', '▁sur', '▁le', '▁bord', '▁de', '▁l', "'", 'extinction', '▁parce', '▁que', '▁ce', '▁n', "'", 'est', '▁pas', '▁vrai', ',', '▁et', '▁pourtant', ',', '▁le', '▁scénario', '▁le', '▁plus', '▁a', 'p', 'ly', 'p', 'é', 'n', 'ique', '▁dans', '▁le', '▁domaine', '▁de', '▁la', '▁diversité', '▁biologique', '▁--', '▁des', '▁approche', 's', '▁rare', 'ment', '▁ce', '▁que', '▁nous', '▁savons', '▁être', '▁le', '▁scénario', '▁le', '▁plus', '▁optimiste', '▁dans', '▁le', '▁domaine', '▁de', '▁la', '▁diversité', '▁culturelle', '.'] 2024-01-16 12:00:36,747 - INFO - joeynmt.training - Epoch 4, Step: 9100, Batch Loss: 1.684603, Batch Acc: 0.617578, Tokens per Sec: 16444, Lr: 0.000182 2024-01-16 12:01:53,532 - INFO - joeynmt.training - Epoch 4, Step: 9200, Batch Loss: 1.795533, Batch Acc: 0.617091, Tokens per Sec: 16231, Lr: 0.000184 2024-01-16 12:03:10,247 - INFO - joeynmt.training - Epoch 4, Step: 9300, Batch Loss: 1.712371, Batch Acc: 0.617977, Tokens per Sec: 16305, Lr: 0.000186 2024-01-16 12:04:27,312 - INFO - joeynmt.training - Epoch 4, Step: 9400, Batch Loss: 1.742355, Batch Acc: 0.619103, Tokens per Sec: 16293, Lr: 0.000188 2024-01-16 12:05:43,623 - INFO - joeynmt.training - Epoch 4, Step: 9500, Batch Loss: 1.621306, Batch Acc: 0.620080, Tokens per Sec: 16385, Lr: 0.000190 2024-01-16 12:07:01,210 - INFO - joeynmt.training - Epoch 4, Step: 9600, Batch Loss: 1.737939, Batch Acc: 0.621094, Tokens per Sec: 16187, Lr: 0.000192 2024-01-16 12:08:17,661 - INFO - joeynmt.training - Epoch 4, Step: 9700, Batch Loss: 1.724599, Batch Acc: 0.623335, Tokens per Sec: 16365, Lr: 0.000194 2024-01-16 12:09:34,522 - INFO - joeynmt.training - Epoch 4, Step: 9800, Batch Loss: 1.697930, Batch Acc: 0.624293, Tokens per Sec: 16327, Lr: 0.000196 2024-01-16 12:10:51,365 - INFO - joeynmt.training - Epoch 4, Step: 9900, Batch Loss: 1.705579, Batch Acc: 0.625993, Tokens per Sec: 16181, Lr: 0.000198 2024-01-16 12:12:08,293 - INFO - joeynmt.training - Epoch 4, Step: 10000, Batch Loss: 1.656694, Batch Acc: 0.627362, Tokens per Sec: 16115, Lr: 0.000200 2024-01-16 12:12:08,294 - INFO - joeynmt.datasets - Sample random subset from dev data: n=500, seed=10042 2024-01-16 12:12:08,295 - INFO - joeynmt.prediction - Predicting 500 example(s)... (Greedy decoding with min_output_length=1, max_output_length=512, return_prob='none', generate_unk=True, repetition_penalty=-1, no_repeat_ngram_size=-1) 2024-01-16 12:12:23,272 - INFO - joeynmt.prediction - Generation took 14.9694[sec]. 2024-01-16 12:12:23,360 - INFO - joeynmt.metrics - nrefs:1|case:lc|eff:no|tok:13a|smooth:exp|version:2.4.0 2024-01-16 12:12:23,361 - INFO - joeynmt.prediction - Evaluation result (greedy): bleu: 23.90, loss: 2.33, ppl: 10.23, acc: 0.53, 0.0709[sec] 2024-01-16 12:12:23,361 - INFO - joeynmt.training - Hooray! New best validation result [bleu]! 2024-01-16 12:12:26,341 - INFO - joeynmt.training - Checkpoint saved in models/iwslt14_prompt/10000.ckpt. 2024-01-16 12:12:26,758 - INFO - joeynmt.training - Example #0 2024-01-16 12:12:26,759 - INFO - joeynmt.training - Source: Le fait de savoir que les Jaguar shaman voyagent toujours au-delà de la voie lactée, ou que les mythes des anciens Inuit résonnent encore de sens, ou bien que dans l'Himalaya, les Bouddhistes continuent à rechercher le souffle du Dharma, c'est se rappeler de la révélation essentielle de l'anthropologie, et cela veut dire que le monde dans lequel nous vivons n'existe pas dans un sens absolu, mais est uniquement un exemple de réalité, la conséquence d'un ensemble spécifique de choix adaptés établis par notre lignée avec succès, il y a plusieurs générations. 2024-01-16 12:12:26,759 - INFO - joeynmt.training - Reference: Just to know that Jaguar shamans still journey beyond the Milky Way, or the myths of the Inuit elders still resonate with meaning, or that in the Himalaya, the Buddhists still pursue the breath of the Dharma, is to really remember the central revelation of anthropology, and that is the idea that the world in which we live does not exist in some absolute sense, but is just one model of reality, the consequence of one particular set of adaptive choices that our lineage made, albeit successfully, many generations ago. 2024-01-16 12:12:26,759 - INFO - joeynmt.training - Hypothesis: The fact that Jaguar Chatt has always traveled beyond the lace, or that ancient Inuit myths still reflect, or even in the Himalayas, the Boudhist continues to look for the breath of the Dharma, is to remember the essential revelation of theanthropology, and it means that the world that in which we live in a way of the world is not only a consequence of our treatment with decades ago. 2024-01-16 12:12:26,761 - DEBUG - joeynmt.training - Tokenized source: ['▁Le', '▁fait', '▁de', '▁savoir', '▁que', '▁les', '▁Ja', 'gu', 'ar', '▁shaman', '▁voyage', 'nt', '▁toujours', '▁au', '-', 'delà', '▁de', '▁la', '▁voie', '▁l', 'act', 'ée', ',', '▁ou', '▁que', '▁les', '▁mythe', 's', '▁des', '▁anciens', '▁Inuit', '▁résonne', 'nt', '▁encore', '▁de', '▁sens', ',', '▁ou', '▁bien', '▁que', '▁dans', '▁l', "'", 'Himalaya', ',', '▁les', '▁Bo', 'ud', 'd', 'h', 'istes', '▁continuent', '▁à', '▁recherche', 'r', '▁le', '▁souffle', '▁du', '▁D', 'ha', 'r', 'ma', ',', '▁c', "'", 'est', '▁se', '▁rappeler', '▁de', '▁la', '▁révélation', '▁essentielle', '▁de', '▁l', "'", 'anthrop', 'ologie', ',', '▁et', '▁cela', '▁veut', '▁dire', '▁que', '▁le', '▁monde', '▁dans', '▁lequel', '▁nous', '▁vivons', '▁n', "'", 'existe', '▁pas', '▁dans', '▁un', '▁sens', '▁absolu', ',', '▁mais', '▁est', '▁uniquement', '▁un', '▁exemple', '▁de', '▁réalité', ',', '▁la', '▁conséquence', '▁d', "'", 'un', '▁ensemble', '▁spécifique', '▁de', '▁choix', '▁adapté', 's', '▁établi', 's', '▁par', '▁notre', '▁lignée', '▁avec', '▁succès', ',', '▁il', '▁y', '▁a', '▁plusieurs', '▁générations', '.'] 2024-01-16 12:12:26,761 - DEBUG - joeynmt.training - Tokenized reference: ['▁Just', '▁to', '▁know', '▁that', '▁Ja', 'gu', 'ar', '▁shaman', 's', '▁still', '▁journey', '▁beyond', '▁the', '▁Milk', 'y', '▁Way', ',', '▁or', '▁the', '▁myth', 's', '▁of', '▁the', '▁Inuit', '▁el', 'der', 's', '▁still', '▁resonate', '▁with', '▁meaning', ',', '▁or', '▁that', '▁in', '▁the', '▁Himalaya', ',', '▁the', '▁Buddhist', 's', '▁still', '▁pursue', '▁the', '▁breath', '▁of', '▁the', '▁D', 'ha', 'r', 'ma', ',', '▁is', '▁to', '▁really', '▁remember', '▁the', '▁central', '▁revelation', '▁of', '▁anthropology', ',', '▁and', '▁that', '▁is', '▁the', '▁idea', '▁that', '▁the', '▁world', '▁in', '▁which', '▁we', '▁live', '▁does', '▁not', '▁exist', '▁in', '▁some', '▁absolute', '▁sense', ',', '▁but', '▁is', '▁just', '▁one', '▁model', '▁of', '▁reality', ',', '▁the', '▁consequence', '▁of', '▁one', '▁particular', '▁set', '▁of', '▁adaptive', '▁choices', '▁that', '▁our', '▁line', 'age', '▁made', ',', '▁albeit', '▁successfully', ',', '▁many', '▁generations', '▁ago', '.'] 2024-01-16 12:12:26,761 - DEBUG - joeynmt.training - Tokenized hypothesis: ['', '▁You', '▁know', ',', '▁one', '▁of', '▁the', '▁intense', '▁pleasure', 's', '▁of', '▁travel', '▁and', '▁one', '▁of', '▁the', '▁de', 'light', 's', '▁of', '▁et', 'h', 'n', 'ographic', '▁research', '▁is', '▁the', '▁opportunity', '▁to', '▁live', '▁among', 'st', '▁those', '▁who', '▁have', '▁not', '▁forgotten', '▁the', '▁old', '▁ways', ',', '▁who', '▁still', '▁feel', '▁their', '▁past', '▁in', '▁the', '▁wind', ',', '▁touch', '▁it', '▁in', '▁stones', '▁polish', 'ed', '▁by', '▁rain', ',', '▁taste', '▁it', '▁in', '▁the', '▁bitter', '▁leaves', '▁of', '▁plants', '.', '', '▁The', '▁fact', '▁that', '▁Ja', 'gu', 'ar', '▁Cha', 't', 't', '▁has', '▁always', '▁traveled', '▁beyond', '▁the', '▁la', 'ce', ',', '▁or', '▁that', '▁ancient', '▁Inuit', '▁myth', 's', '▁still', '▁reflect', ',', '▁or', '▁even', '▁in', '▁the', '▁Himalaya', 's', ',', '▁the', '▁Bo', 'ud', 'h', 'ist', '▁continues', '▁to', '▁look', '▁for', '▁the', '▁breath', '▁of', '▁the', '▁D', 'ha', 'r', 'ma', ',', '▁is', '▁to', '▁remember', '▁the', '▁essential', '▁revelation', '▁of', '▁the', 'anthrop', 'ology', ',', '▁and', '▁it', '▁means', '▁that', '▁the', '▁world', '▁that', '▁in', '▁which', '▁we', '▁live', '▁in', '▁a', '▁way', '▁of', '▁the', '▁world', '▁is', '▁not', '▁only', '▁a', '▁consequence', '▁of', '▁our', '▁treatment', '▁with', '▁decades', '▁ago', '.'] 2024-01-16 12:12:26,761 - INFO - joeynmt.training - Example #1 2024-01-16 12:12:26,761 - INFO - joeynmt.training - Source: We're all born. We all bring our children into the world. 2024-01-16 12:12:26,761 - INFO - joeynmt.training - Reference: Nous sommes tous nés. Nous apportons nos enfants dans ce monde. 2024-01-16 12:12:26,761 - INFO - joeynmt.training - Hypothesis: Nous sommes tous nés. Nous nous nous sommes tous tous en train de nous amener nos enfants dans le monde. 2024-01-16 12:12:26,762 - DEBUG - joeynmt.training - Tokenized source: ['▁We', "'", 're', '▁all', '▁born', '.', '▁We', '▁all', '▁bring', '▁our', '▁children', '▁into', '▁the', '▁world', '.'] 2024-01-16 12:12:26,762 - DEBUG - joeynmt.training - Tokenized reference: ['▁Nous', '▁sommes', '▁tous', '▁né', 's', '.', '▁Nous', '▁apport', 'ons', '▁nos', '▁enfants', '▁dans', '▁ce', '▁monde', '.'] 2024-01-16 12:12:26,762 - DEBUG - joeynmt.training - Tokenized hypothesis: ['', '▁Bien', '▁sûr', ',', '▁nous', '▁partageons', '▁tous', '▁les', '▁mêmes', '▁im', 'pé', 'r', 'atif', 's', '▁d', "'", 'adaptation', '.', '', '▁Nous', '▁sommes', '▁tous', '▁né', 's', '.', '▁Nous', '▁nous', '▁nous', '▁sommes', '▁tous', '▁tous', '▁en', '▁train', '▁de', '▁nous', '▁amener', '▁nos', '▁enfants', '▁dans', '▁le', '▁monde', '.'] 2024-01-16 12:12:26,762 - INFO - joeynmt.training - Example #2 2024-01-16 12:12:26,763 - INFO - joeynmt.training - Source: It's the symbol of all that we are and all that we can be as an astonishingly inquisitive species. 2024-01-16 12:12:26,763 - INFO - joeynmt.training - Reference: C'est le symbole de tout ce que nous sommes et tout ce que nous pouvons être en tant qu'espèce dotée d'une curiosité stupéfiante. 2024-01-16 12:12:26,763 - INFO - joeynmt.training - Hypothesis: C'est le symbole de tout ce que nous sommes et tout ce que nous pouvons être comme une espèce étonnantement inquisitive. 2024-01-16 12:12:26,764 - DEBUG - joeynmt.training - Tokenized source: ['▁It', "'", 's', '▁the', '▁symbol', '▁of', '▁all', '▁that', '▁we', '▁are', '▁and', '▁all', '▁that', '▁we', '▁can', '▁be', '▁as', '▁an', '▁astonishing', 'ly', '▁in', 'qui', 's', 'it', 'ive', '▁species', '.'] 2024-01-16 12:12:26,764 - DEBUG - joeynmt.training - Tokenized reference: ['▁C', "'", 'est', '▁le', '▁symbole', '▁de', '▁tout', '▁ce', '▁que', '▁nous', '▁sommes', '▁et', '▁tout', '▁ce', '▁que', '▁nous', '▁pouvons', '▁être', '▁en', '▁tant', '▁qu', "'", 'espèce', '▁do', 't', 'ée', '▁d', "'", 'une', '▁curiosité', '▁stupéfiant', 'e', '.'] 2024-01-16 12:12:26,764 - DEBUG - joeynmt.training - Tokenized hypothesis: ['', '▁L', "'", 'e', 'th', 'no', 's', 'ph', 'ère', '▁est', '▁l', "'", 'héritage', '▁de', '▁l', "'", 'humanité', '.', '', '▁C', "'", 'est', '▁le', '▁symbole', '▁de', '▁tout', '▁ce', '▁que', '▁nous', '▁sommes', '▁et', '▁tout', '▁ce', '▁que', '▁nous', '▁pouvons', '▁être', '▁comme', '▁une', '▁espèce', '▁étonnante', 'ment', '▁in', 'qui', 's', 'it', 'ive', '.'] 2024-01-16 12:12:26,764 - INFO - joeynmt.training - Example #3 2024-01-16 12:12:26,764 - INFO - joeynmt.training - Source: No biologists, for example, would dare suggest that 50 percent of all species or more have been or are on the brink of extinction because it simply is not true, and yet that -- the most apocalyptic scenario in the realm of biological diversity -- scarcely approaches what we know to be the most optimistic scenario in the realm of cultural diversity. 2024-01-16 12:12:26,764 - INFO - joeynmt.training - Reference: Aucun biologiste, par exemple, n'oserait suggérer que 50% ou plus de toutes les espèces ont été ou sont à deux doigts de l'extinction parce que ce n'est tout simplement pas vrai, et pourtant -- que le scénario le plus apocalyptique dans le royaume de la diversité biologique -- se rapproche rarement de ce que nous considérons comme le scénario le plus optimiste au sein de la diversité culturelle. 2024-01-16 12:12:26,764 - INFO - joeynmt.training - Hypothesis: Les biologistes, par exemple, s'y suggéreraient que 50 % de toutes les espèces ou plus ont été ou sont sur le bord de l'extinction parce que ce n'est pas vrai, et pourtant que -- le scénario le plus apployé dans le domaine de la diversité biologique -- des approches rarement ce que nous savons être le scénario le plus optimiste dans le domaine de la diversité culturelle. 2024-01-16 12:12:26,765 - DEBUG - joeynmt.training - Tokenized source: ['▁No', '▁biologists', ',', '▁for', '▁example', ',', '▁would', '▁da', 're', '▁suggest', '▁that', '▁50', '▁percent', '▁of', '▁all', '▁species', '▁or', '▁more', '▁have', '▁been', '▁or', '▁are', '▁on', '▁the', '▁brin', 'k', '▁of', '▁extinction', '▁because', '▁it', '▁simply', '▁is', '▁not', '▁true', ',', '▁and', '▁yet', '▁that', '▁--', '▁the', '▁most', '▁apocalyptic', '▁scenario', '▁in', '▁the', '▁realm', '▁of', '▁biological', '▁diversity', '▁--', '▁scarce', 'ly', '▁approaches', '▁what', '▁we', '▁know', '▁to', '▁be', '▁the', '▁most', '▁optimistic', '▁scenario', '▁in', '▁the', '▁realm', '▁of', '▁cultural', '▁diversity', '.'] 2024-01-16 12:12:26,765 - DEBUG - joeynmt.training - Tokenized reference: ['▁Aucun', '▁biologist', 'e', ',', '▁par', '▁exemple', ',', '▁n', "'", 'ose', 'rait', '▁suggérer', '▁que', '▁50%', '▁ou', '▁plus', '▁de', '▁toutes', '▁les', '▁espèces', '▁ont', '▁été', '▁ou', '▁sont', '▁à', '▁deux', '▁doigts', '▁de', '▁l', "'", 'extinction', '▁parce', '▁que', '▁ce', '▁n', "'", 'est', '▁tout', '▁simplement', '▁pas', '▁vrai', ',', '▁et', '▁pourtant', '▁--', '▁que', '▁le', '▁scénario', '▁le', '▁plus', '▁a', 'po', 'ca', 'ly', 'p', 'tique', '▁dans', '▁le', '▁royaume', '▁de', '▁la', '▁diversité', '▁biologique', '▁--', '▁se', '▁rapproche', '▁rarement', '▁de', '▁ce', '▁que', '▁nous', '▁considér', 'ons', '▁comme', '▁le', '▁scénario', '▁le', '▁plus', '▁optimiste', '▁au', '▁sein', '▁de', '▁la', '▁diversité', '▁culturelle', '.'] 2024-01-16 12:12:26,765 - DEBUG - joeynmt.training - Tokenized hypothesis: ['', '▁Et', '▁lorsque', '▁la', '▁biosphère', '▁fut', '▁sérieusement', '▁é', 'ro', 'd', 'ée', ',', '▁l', "'", 'e', 'th', 'no', 's', 'ph', 'ère', '▁l', "'", 'a', '▁été', '▁également', '▁--', '▁et', '▁peut', '-', 'être', '▁bien', '▁plus', '▁rapidement', '.', '', '▁Les', '▁biologistes', ',', '▁par', '▁exemple', ',', '▁s', "'", 'y', '▁suggérer', 'aient', '▁que', '▁50', '▁%', '▁de', '▁toutes', '▁les', '▁espèces', '▁ou', '▁plus', '▁ont', '▁été', '▁ou', '▁sont', '▁sur', '▁le', '▁bord', '▁de', '▁l', "'", 'extinction', '▁parce', '▁que', '▁ce', '▁n', "'", 'est', '▁pas', '▁vrai', ',', '▁et', '▁pourtant', '▁que', '▁--', '▁le', '▁scénario', '▁le', '▁plus', '▁a', 'p', 'p', 'lo', 'y', 'é', '▁dans', '▁le', '▁domaine', '▁de', '▁la', '▁diversité', '▁biologique', '▁--', '▁des', '▁approche', 's', '▁rare', 'ment', '▁ce', '▁que', '▁nous', '▁savons', '▁être', '▁le', '▁scénario', '▁le', '▁plus', '▁optimiste', '▁dans', '▁le', '▁domaine', '▁de', '▁la', '▁diversité', '▁culturelle', '.'] 2024-01-16 12:13:43,771 - INFO - joeynmt.training - Epoch 4, Step: 10100, Batch Loss: 1.666885, Batch Acc: 0.628234, Tokens per Sec: 16265, Lr: 0.000199 2024-01-16 12:15:00,288 - INFO - joeynmt.training - Epoch 4, Step: 10200, Batch Loss: 1.688828, Batch Acc: 0.629902, Tokens per Sec: 16334, Lr: 0.000198 2024-01-16 12:16:16,984 - INFO - joeynmt.training - Epoch 4, Step: 10300, Batch Loss: 1.641827, Batch Acc: 0.631982, Tokens per Sec: 16240, Lr: 0.000197 2024-01-16 12:17:33,629 - INFO - joeynmt.training - Epoch 4, Step: 10400, Batch Loss: 1.656813, Batch Acc: 0.630895, Tokens per Sec: 16276, Lr: 0.000196 2024-01-16 12:18:52,615 - INFO - joeynmt.training - Epoch 4, Step: 10500, Batch Loss: 1.628379, Batch Acc: 0.633568, Tokens per Sec: 15839, Lr: 0.000195 2024-01-16 12:20:08,791 - INFO - joeynmt.training - Epoch 4, Step: 10600, Batch Loss: 1.662849, Batch Acc: 0.635515, Tokens per Sec: 16336, Lr: 0.000194 2024-01-16 12:21:25,126 - INFO - joeynmt.training - Epoch 4, Step: 10700, Batch Loss: 1.685087, Batch Acc: 0.635510, Tokens per Sec: 16341, Lr: 0.000193 2024-01-16 12:22:41,269 - INFO - joeynmt.training - Epoch 4, Step: 10800, Batch Loss: 1.616188, Batch Acc: 0.637787, Tokens per Sec: 16339, Lr: 0.000192 2024-01-16 12:23:58,756 - INFO - joeynmt.training - Epoch 4, Step: 10900, Batch Loss: 1.595238, Batch Acc: 0.638838, Tokens per Sec: 16112, Lr: 0.000192 2024-01-16 12:24:54,560 - INFO - joeynmt.training - Epoch 4, total training loss: 4707.17, num. of seqs: 702202, num. of tokens: 34266040, 2106.8837[sec] 2024-01-16 12:24:54,570 - INFO - joeynmt.training - EPOCH 5 2024-01-16 12:25:16,304 - INFO - joeynmt.training - Epoch 5, Step: 11000, Batch Loss: 1.476027, Batch Acc: 0.653803, Tokens per Sec: 16100, Lr: 0.000191 2024-01-16 12:25:16,305 - INFO - joeynmt.datasets - Sample random subset from dev data: n=500, seed=11042 2024-01-16 12:25:16,305 - INFO - joeynmt.prediction - Predicting 500 example(s)... (Greedy decoding with min_output_length=1, max_output_length=512, return_prob='none', generate_unk=True, repetition_penalty=-1, no_repeat_ngram_size=-1) 2024-01-16 12:25:30,277 - INFO - joeynmt.prediction - Generation took 13.9639[sec]. 2024-01-16 12:25:30,433 - INFO - joeynmt.metrics - nrefs:1|case:lc|eff:no|tok:13a|smooth:exp|version:2.4.0 2024-01-16 12:25:30,433 - INFO - joeynmt.prediction - Evaluation result (greedy): bleu: 24.48, loss: 2.27, ppl: 9.66, acc: 0.55, 0.1179[sec] 2024-01-16 12:25:30,434 - INFO - joeynmt.training - Hooray! New best validation result [bleu]! 2024-01-16 12:25:33,534 - INFO - joeynmt.training - Checkpoint saved in models/iwslt14_prompt/11000.ckpt. 2024-01-16 12:25:33,537 - INFO - joeynmt.training - Example #0 2024-01-16 12:25:33,538 - INFO - joeynmt.training - Source: You know, one of the intense pleasures of travel and one of the delights of ethnographic research is the opportunity to live amongst those who have not forgotten the old ways, who still feel their past in the wind, touch it in stones polished by rain, taste it in the bitter leaves of plants. 2024-01-16 12:25:33,538 - INFO - joeynmt.training - Reference: Vous savez, un des plaisirs intenses du voyage et un des délices de la recherche ethnographique est la possibilité de vivre parmi ceux qui n'ont pas oublié les anciennes coutumes, qui ressentent encore leur passé souffler dans le vent, qui le touchent dans les pierres polies par la pluie, le dégustent dans les feuilles amères des plantes. 2024-01-16 12:25:33,538 - INFO - joeynmt.training - Hypothesis: Vous savez, l'un des plaisirs intenses de voyage et l'une des plaisirs de la recherche ethnographic est l'opportunité de vivre parmi ceux qui n'ont pas oublié les vieilles manières, qui se sent encore au vent, qui le touchait en pierre, qui l'a fait en colère par la pluie, le goût dans les lars des plantes. 2024-01-16 12:25:33,539 - DEBUG - joeynmt.training - Tokenized source: ['▁You', '▁know', ',', '▁one', '▁of', '▁the', '▁intense', '▁pleasure', 's', '▁of', '▁travel', '▁and', '▁one', '▁of', '▁the', '▁de', 'light', 's', '▁of', '▁et', 'h', 'n', 'ographic', '▁research', '▁is', '▁the', '▁opportunity', '▁to', '▁live', '▁among', 'st', '▁those', '▁who', '▁have', '▁not', '▁forgotten', '▁the', '▁old', '▁ways', ',', '▁who', '▁still', '▁feel', '▁their', '▁past', '▁in', '▁the', '▁wind', ',', '▁touch', '▁it', '▁in', '▁stones', '▁polish', 'ed', '▁by', '▁rain', ',', '▁taste', '▁it', '▁in', '▁the', '▁bitter', '▁leaves', '▁of', '▁plants', '.'] 2024-01-16 12:25:33,539 - DEBUG - joeynmt.training - Tokenized reference: ['▁Vous', '▁savez', ',', '▁un', '▁des', '▁plaisir', 's', '▁intense', 's', '▁du', '▁voyage', '▁et', '▁un', '▁des', '▁dé', 'lic', 'es', '▁de', '▁la', '▁recherche', '▁et', 'h', 'no', 'graph', 'ique', '▁est', '▁la', '▁possibilité', '▁de', '▁vivre', '▁parmi', '▁ceux', '▁qui', '▁n', "'", 'ont', '▁pas', '▁oublié', '▁les', '▁anciennes', '▁coutume', 's', ',', '▁qui', '▁ressentent', '▁encore', '▁leur', '▁passé', '▁souffle', 'r', '▁dans', '▁le', '▁vent', ',', '▁qui', '▁le', '▁touchent', '▁dans', '▁les', '▁pierres', '▁poli', 'es', '▁par', '▁la', '▁pluie', ',', '▁le', '▁dé', 'gu', 'sten', 't', '▁dans', '▁les', '▁feuilles', '▁a', 'mère', 's', '▁des', '▁plantes', '.'] 2024-01-16 12:25:33,539 - DEBUG - joeynmt.training - Tokenized hypothesis: ['', '', '▁Vous', '▁savez', ',', '▁l', "'", 'un', '▁des', '▁plaisir', 's', '▁intense', 's', '▁de', '▁voyage', '▁et', '▁l', "'", 'une', '▁des', '▁plaisir', 's', '▁de', '▁la', '▁recherche', '▁et', 'h', 'n', 'ographic', '▁est', '▁l', "'", 'opportunité', '▁de', '▁vivre', '▁parmi', '▁ceux', '▁qui', '▁n', "'", 'ont', '▁pas', '▁oublié', '▁les', '▁vieilles', '▁manières', ',', '▁qui', '▁se', '▁sent', '▁encore', '▁au', '▁vent', ',', '▁qui', '▁le', '▁touch', 'ait', '▁en', '▁pierre', ',', '▁qui', '▁l', "'", 'a', '▁fait', '▁en', '▁colère', '▁par', '▁la', '▁pluie', ',', '▁le', '▁goût', '▁dans', '▁les', '▁la', 'r', 's', '▁des', '▁plantes', '.'] 2024-01-16 12:25:33,539 - INFO - joeynmt.training - Example #1 2024-01-16 12:25:33,539 - INFO - joeynmt.training - Source: C'est le symbole de tout ce que nous sommes et tout ce que nous pouvons être en tant qu'espèce dotée d'une curiosité stupéfiante. 2024-01-16 12:25:33,539 - INFO - joeynmt.training - Reference: It's the symbol of all that we are and all that we can be as an astonishingly inquisitive species. 2024-01-16 12:25:33,540 - INFO - joeynmt.training - Hypothesis: It's the symbol of all we are and all we can be as a species of amazing curiosity. 2024-01-16 12:25:33,540 - DEBUG - joeynmt.training - Tokenized source: ['▁C', "'", 'est', '▁le', '▁symbole', '▁de', '▁tout', '▁ce', '▁que', '▁nous', '▁sommes', '▁et', '▁tout', '▁ce', '▁que', '▁nous', '▁pouvons', '▁être', '▁en', '▁tant', '▁qu', "'", 'espèce', '▁do', 't', 'ée', '▁d', "'", 'une', '▁curiosité', '▁stupéfiant', 'e', '.'] 2024-01-16 12:25:33,540 - DEBUG - joeynmt.training - Tokenized reference: ['▁It', "'", 's', '▁the', '▁symbol', '▁of', '▁all', '▁that', '▁we', '▁are', '▁and', '▁all', '▁that', '▁we', '▁can', '▁be', '▁as', '▁an', '▁astonishing', 'ly', '▁in', 'qui', 's', 'it', 'ive', '▁species', '.'] 2024-01-16 12:25:33,540 - DEBUG - joeynmt.training - Tokenized hypothesis: ['', '▁The', '▁et', 'h', 'n', 'osphere', '▁is', '▁humanity', "'", 's', '▁great', '▁legacy', '.', '', '▁It', "'", 's', '▁the', '▁symbol', '▁of', '▁all', '▁we', '▁are', '▁and', '▁all', '▁we', '▁can', '▁be', '▁as', '▁a', '▁species', '▁of', '▁amazing', '▁curiosity', '.'] 2024-01-16 12:25:33,540 - INFO - joeynmt.training - Example #2 2024-01-16 12:25:33,541 - INFO - joeynmt.training - Source: Aucun biologiste, par exemple, n'oserait suggérer que 50% ou plus de toutes les espèces ont été ou sont à deux doigts de l'extinction parce que ce n'est tout simplement pas vrai, et pourtant -- que le scénario le plus apocalyptique dans le royaume de la diversité biologique -- se rapproche rarement de ce que nous considérons comme le scénario le plus optimiste au sein de la diversité culturelle. 2024-01-16 12:25:33,541 - INFO - joeynmt.training - Reference: No biologists, for example, would dare suggest that 50 percent of all species or more have been or are on the brink of extinction because it simply is not true, and yet that -- the most apocalyptic scenario in the realm of biological diversity -- scarcely approaches what we know to be the most optimistic scenario in the realm of cultural diversity. 2024-01-16 12:25:33,541 - INFO - joeynmt.training - Hypothesis: No biologist, for example, would suggest that 50 or more species have been or two fingers of extinction because it's not just true, and yet -- the most abnormal scenario in the realm of biological diversity -- rarely seem to be closer to what we consider the most optimistic scenario in the breast of cultural diversity. 2024-01-16 12:25:33,542 - DEBUG - joeynmt.training - Tokenized source: ['▁Aucun', '▁biologist', 'e', ',', '▁par', '▁exemple', ',', '▁n', "'", 'ose', 'rait', '▁suggérer', '▁que', '▁50%', '▁ou', '▁plus', '▁de', '▁toutes', '▁les', '▁espèces', '▁ont', '▁été', '▁ou', '▁sont', '▁à', '▁deux', '▁doigts', '▁de', '▁l', "'", 'extinction', '▁parce', '▁que', '▁ce', '▁n', "'", 'est', '▁tout', '▁simplement', '▁pas', '▁vrai', ',', '▁et', '▁pourtant', '▁--', '▁que', '▁le', '▁scénario', '▁le', '▁plus', '▁a', 'po', 'ca', 'ly', 'p', 'tique', '▁dans', '▁le', '▁royaume', '▁de', '▁la', '▁diversité', '▁biologique', '▁--', '▁se', '▁rapproche', '▁rarement', '▁de', '▁ce', '▁que', '▁nous', '▁considér', 'ons', '▁comme', '▁le', '▁scénario', '▁le', '▁plus', '▁optimiste', '▁au', '▁sein', '▁de', '▁la', '▁diversité', '▁culturelle', '.'] 2024-01-16 12:25:33,542 - DEBUG - joeynmt.training - Tokenized reference: ['▁No', '▁biologists', ',', '▁for', '▁example', ',', '▁would', '▁da', 're', '▁suggest', '▁that', '▁50', '▁percent', '▁of', '▁all', '▁species', '▁or', '▁more', '▁have', '▁been', '▁or', '▁are', '▁on', '▁the', '▁brin', 'k', '▁of', '▁extinction', '▁because', '▁it', '▁simply', '▁is', '▁not', '▁true', ',', '▁and', '▁yet', '▁that', '▁--', '▁the', '▁most', '▁apocalyptic', '▁scenario', '▁in', '▁the', '▁realm', '▁of', '▁biological', '▁diversity', '▁--', '▁scarce', 'ly', '▁approaches', '▁what', '▁we', '▁know', '▁to', '▁be', '▁the', '▁most', '▁optimistic', '▁scenario', '▁in', '▁the', '▁realm', '▁of', '▁cultural', '▁diversity', '.'] 2024-01-16 12:25:33,542 - DEBUG - joeynmt.training - Tokenized hypothesis: ['', '▁And', '▁just', '▁as', '▁the', '▁biosphere', '▁has', '▁been', '▁severely', '▁er', 'od', 'ed', ',', '▁so', '▁too', '▁is', '▁the', '▁et', 'h', 'n', 'osphere', '▁--', '▁and', ',', '▁if', '▁anything', ',', '▁at', '▁a', '▁far', '▁greater', '▁rate', '.', '', '▁No', '▁biologist', ',', '▁for', '▁example', ',', '▁would', '▁suggest', '▁that', '▁50', '▁or', '▁more', '▁species', '▁have', '▁been', '▁or', '▁two', '▁fingers', '▁of', '▁extinction', '▁because', '▁it', "'", 's', '▁not', '▁just', '▁true', ',', '▁and', '▁yet', '▁--', '▁the', '▁most', '▁abnormal', '▁scenario', '▁in', '▁the', '▁realm', '▁of', '▁biological', '▁diversity', '▁--', '▁rarely', '▁seem', '▁to', '▁be', '▁closer', '▁to', '▁what', '▁we', '▁consider', '▁the', '▁most', '▁optimistic', '▁scenario', '▁in', '▁the', '▁breast', '▁of', '▁cultural', '▁diversity', '.'] 2024-01-16 12:25:33,542 - INFO - joeynmt.training - Example #3 2024-01-16 12:25:33,542 - INFO - joeynmt.training - Source: Et l'indicateur le plus fiable est bien sûr l'extinction du langage. 2024-01-16 12:25:33,542 - INFO - joeynmt.training - Reference: And the great indicator of that, of course, is language loss. 2024-01-16 12:25:33,542 - INFO - joeynmt.training - Hypothesis: And the most reliable indicator is of course the extinction of language. 2024-01-16 12:25:33,543 - DEBUG - joeynmt.training - Tokenized source: ['▁Et', '▁l', "'", 'ind', 'ic', 'ateur', '▁le', '▁plus', '▁fiable', '▁est', '▁bien', '▁sûr', '▁l', "'", 'extinction', '▁du', '▁langage', '.'] 2024-01-16 12:25:33,543 - DEBUG - joeynmt.training - Tokenized reference: ['▁And', '▁the', '▁great', '▁indicator', '▁of', '▁that', ',', '▁of', '▁course', ',', '▁is', '▁language', '▁loss', '.'] 2024-01-16 12:25:33,543 - DEBUG - joeynmt.training - Tokenized hypothesis: ['', '▁No', '▁biologists', ',', '▁for', '▁example', ',', '▁would', '▁da', 're', '▁suggest', '▁that', '▁50', '▁percent', '▁of', '▁all', '▁species', '▁or', '▁more', '▁have', '▁been', '▁or', '▁are', '▁on', '▁the', '▁brin', 'k', '▁of', '▁extinction', '▁because', '▁it', '▁simply', '▁is', '▁not', '▁true', ',', '▁and', '▁yet', '▁that', '▁--', '▁the', '▁most', '▁apocalyptic', '▁scenario', '▁in', '▁the', '▁realm', '▁of', '▁biological', '▁diversity', '▁--', '▁scarce', 'ly', '▁approaches', '▁what', '▁we', '▁know', '▁to', '▁be', '▁the', '▁most', '▁optimistic', '▁scenario', '▁in', '▁the', '▁realm', '▁of', '▁cultural', '▁diversity', '.', '', '▁And', '▁the', '▁most', '▁reliable', '▁indicator', '▁is', '▁of', '▁course', '▁the', '▁extinction', '▁of', '▁language', '.'] 2024-01-16 12:26:49,376 - INFO - joeynmt.training - Epoch 5, Step: 11100, Batch Loss: 1.489354, Batch Acc: 0.651769, Tokens per Sec: 16298, Lr: 0.000190 2024-01-16 12:28:06,201 - INFO - joeynmt.training - Epoch 5, Step: 11200, Batch Loss: 1.504270, Batch Acc: 0.650713, Tokens per Sec: 16297, Lr: 0.000189 2024-01-16 12:29:22,516 - INFO - joeynmt.training - Epoch 5, Step: 11300, Batch Loss: 1.551262, Batch Acc: 0.652619, Tokens per Sec: 16352, Lr: 0.000188 2024-01-16 12:30:39,489 - INFO - joeynmt.training - Epoch 5, Step: 11400, Batch Loss: 1.515537, Batch Acc: 0.653057, Tokens per Sec: 16225, Lr: 0.000187 2024-01-16 12:31:55,518 - INFO - joeynmt.training - Epoch 5, Step: 11500, Batch Loss: 1.504352, Batch Acc: 0.653733, Tokens per Sec: 16435, Lr: 0.000187 2024-01-16 12:33:11,021 - INFO - joeynmt.training - Epoch 5, Step: 11600, Batch Loss: 1.518028, Batch Acc: 0.654302, Tokens per Sec: 16473, Lr: 0.000186 2024-01-16 12:34:28,406 - INFO - joeynmt.training - Epoch 5, Step: 11700, Batch Loss: 1.558879, Batch Acc: 0.656093, Tokens per Sec: 16132, Lr: 0.000185 2024-01-16 12:35:44,867 - INFO - joeynmt.training - Epoch 5, Step: 11800, Batch Loss: 1.501531, Batch Acc: 0.655743, Tokens per Sec: 16385, Lr: 0.000184 2024-01-16 12:37:01,451 - INFO - joeynmt.training - Epoch 5, Step: 11900, Batch Loss: 1.529770, Batch Acc: 0.657034, Tokens per Sec: 16383, Lr: 0.000183 2024-01-16 12:38:16,783 - INFO - joeynmt.training - Epoch 5, Step: 12000, Batch Loss: 1.549649, Batch Acc: 0.657840, Tokens per Sec: 16558, Lr: 0.000183 2024-01-16 12:38:16,784 - INFO - joeynmt.datasets - Sample random subset from dev data: n=500, seed=12042 2024-01-16 12:38:16,784 - INFO - joeynmt.prediction - Predicting 500 example(s)... (Greedy decoding with min_output_length=1, max_output_length=512, return_prob='none', generate_unk=True, repetition_penalty=-1, no_repeat_ngram_size=-1) 2024-01-16 12:38:44,463 - INFO - joeynmt.prediction - Generation took 27.6703[sec]. 2024-01-16 12:38:44,550 - INFO - joeynmt.metrics - nrefs:1|case:lc|eff:no|tok:13a|smooth:exp|version:2.4.0 2024-01-16 12:38:44,550 - INFO - joeynmt.prediction - Evaluation result (greedy): bleu: 26.33, loss: 2.22, ppl: 9.20, acc: 0.55, 0.0697[sec] 2024-01-16 12:38:44,551 - INFO - joeynmt.training - Hooray! New best validation result [bleu]! 2024-01-16 12:38:47,528 - INFO - joeynmt.training - Checkpoint saved in models/iwslt14_prompt/12000.ckpt. 2024-01-16 12:38:47,858 - INFO - joeynmt.training - Example #0 2024-01-16 12:38:47,859 - INFO - joeynmt.training - Source: Par contre, la cadence exceptionnelle de la chanson est intéressante, le rythme de la danse dans toutes les cultures. 2024-01-16 12:38:47,859 - INFO - joeynmt.training - Reference: But what's interesting is the unique cadence of the song, the rhythm of the dance in every culture. 2024-01-16 12:38:47,859 - INFO - joeynmt.training - Hypothesis: And, against, the remarkable cadence of the song is interesting, the rhythm of the dance in all cultures. 2024-01-16 12:38:47,860 - DEBUG - joeynmt.training - Tokenized source: ['▁Par', '▁contre', ',', '▁la', '▁cadence', '▁exceptionnel', 'le', '▁de', '▁la', '▁chanson', '▁est', '▁intéressante', ',', '▁le', '▁rythme', '▁de', '▁la', '▁dans', 'e', '▁dans', '▁toutes', '▁les', '▁cultures', '.'] 2024-01-16 12:38:47,860 - DEBUG - joeynmt.training - Tokenized reference: ['▁But', '▁what', "'", 's', '▁interesting', '▁is', '▁the', '▁unique', '▁cadence', '▁of', '▁the', '▁song', ',', '▁the', '▁rhythm', '▁of', '▁the', '▁dance', '▁in', '▁every', '▁culture', '.'] 2024-01-16 12:38:47,860 - DEBUG - joeynmt.training - Tokenized hypothesis: ['', '▁We', '▁have', '▁to', '▁deal', '▁with', '▁the', '▁inexorable', '▁separation', '▁of', '▁death', ',', '▁so', '▁it', '▁should', 'n', "'", 't', '▁surprise', '▁us', '▁that', '▁we', '▁all', '▁sing', ',', '▁we', '▁all', '▁dance', ',', '▁we', '▁all', '▁have', '▁art', '.', '', '▁And', ',', '▁against', ',', '▁the', '▁remarkable', '▁cadence', '▁of', '▁the', '▁song', '▁is', '▁interesting', ',', '▁the', '▁rhythm', '▁of', '▁the', '▁dance', '▁in', '▁all', '▁cultures', '.'] 2024-01-16 12:38:47,860 - INFO - joeynmt.training - Example #1 2024-01-16 12:38:47,861 - INFO - joeynmt.training - Source: And whether it is the Penan in the forests of Borneo, or the Voodoo acolytes in Haiti, or the warriors in the Kaisut desert of Northern Kenya, the Curandero in the mountains of the Andes, or a caravanserai in the middle of the Sahara -- this is incidentally the fellow that I traveled into the desert with a month ago -- or indeed a yak herder in the slopes of Qomolangma, Everest, the goddess mother of the world. 2024-01-16 12:38:47,861 - INFO - joeynmt.training - Reference: Et que ce soit le Penan dans les forêts du Bornéo, ou les acolytes Voodoo à Haïti, ou bien les guerriers dans le désert du Kaisut au nord du Kenya, le Curendero dans les montagnes des Andes, ou bien un caravansérail en plein milieu du Sahara. A propos, c'est la personne avec qui j'ai voyagé dans le désert il y un mois, ou effectivement, le gardien de troupeau de Yaks sur les flancs du Qomolangma, l'Everest, la déesse du monde. 2024-01-16 12:38:47,861 - INFO - joeynmt.training - Hypothesis: Et si c'est la Penan dans les forêts de Borneo, ou les acolytes Voodoo en Haïti, ou les guerrets du désert Kaisutut du Nord du Kenya, le Curandero dans les montagnes des Andes, ou une voitureavanserai au milieu du Sahara -- c'est à peu près le gars que j'ai voyagé dans le désert avec un mois -- ou en effet un yak dans les pentes de Qomolangma, de l'Everest, la mère de la Gétinité du monde. 2024-01-16 12:38:47,862 - DEBUG - joeynmt.training - Tokenized source: ['▁And', '▁whether', '▁it', '▁is', '▁the', '▁Pen', 'an', '▁in', '▁the', '▁forests', '▁of', '▁Bo', 'r', 'ne', 'o', ',', '▁or', '▁the', '▁V', 'oodoo', '▁a', 'co', 'ly', 'tes', '▁in', '▁Haiti', ',', '▁or', '▁the', '▁warrior', 's', '▁in', '▁the', '▁K', 'ais', 'ut', '▁desert', '▁of', '▁Northern', '▁Kenya', ',', '▁the', '▁C', 'ura', 'nder', 'o', '▁in', '▁the', '▁mountains', '▁of', '▁the', '▁And', 'es', ',', '▁or', '▁a', '▁car', 'ava', 'n', 's', 'er', 'ai', '▁in', '▁the', '▁middle', '▁of', '▁the', '▁Sahara', '▁--', '▁this', '▁is', '▁incidentally', '▁the', '▁fellow', '▁that', '▁I', '▁traveled', '▁into', '▁the', '▁desert', '▁with', '▁a', '▁month', '▁ago', '▁--', '▁or', '▁indeed', '▁a', '▁y', 'ak', '▁her', 'der', '▁in', '▁the', '▁slope', 's', '▁of', '▁Q', 'o', 'mo', 'lang', 'ma', ',', '▁Everest', ',', '▁the', '▁goddess', '▁mother', '▁of', '▁the', '▁world', '.'] 2024-01-16 12:38:47,862 - DEBUG - joeynmt.training - Tokenized reference: ['▁Et', '▁que', '▁ce', '▁soit', '▁le', '▁Pen', 'an', '▁dans', '▁les', '▁forêts', '▁du', '▁Bo', 'r', 'né', 'o', ',', '▁ou', '▁les', '▁a', 'co', 'ly', 'tes', '▁V', 'oodoo', '▁à', '▁Haïti', ',', '▁ou', '▁bien', '▁les', '▁guerrier', 's', '▁dans', '▁le', '▁désert', '▁du', '▁K', 'ais', 'ut', '▁au', '▁nord', '▁du', '▁Kenya', ',', '▁le', '▁C', 'ur', 'ende', 'ro', '▁dans', '▁les', '▁montagnes', '▁des', '▁And', 'es', ',', '▁ou', '▁bien', '▁un', '▁car', 'ava', 'n', 's', 'é', 'ra', 'il', '▁en', '▁plein', '▁milieu', '▁du', '▁Sahara', '.', '▁A', '▁propos', ',', '▁c', "'", 'est', '▁la', '▁personne', '▁avec', '▁qui', '▁j', "'", 'ai', '▁voyagé', '▁dans', '▁le', '▁désert', '▁il', '▁y', '▁un', '▁mois', ',', '▁ou', '▁effectivement', ',', '▁le', '▁gardien', '▁de', '▁troupeau', '▁de', '▁Yak', 's', '▁sur', '▁les', '▁fla', 'n', 'c', 's', '▁du', '▁Q', 'o', 'mo', 'lang', 'ma', ',', '▁l', "'", 'Everest', ',', '▁la', '▁dé', 'esse', '▁du', '▁monde', '.'] 2024-01-16 12:38:47,862 - DEBUG - joeynmt.training - Tokenized hypothesis: ['', '▁Par', '▁contre', ',', '▁la', '▁cadence', '▁exceptionnel', 'le', '▁de', '▁la', '▁chanson', '▁est', '▁intéressante', ',', '▁le', '▁rythme', '▁de', '▁la', '▁dans', 'e', '▁dans', '▁toutes', '▁les', '▁cultures', '.', '', '▁Et', '▁si', '▁c', "'", 'est', '▁la', '▁Pen', 'an', '▁dans', '▁les', '▁forêts', '▁de', '▁Bo', 'r', 'ne', 'o', ',', '▁ou', '▁les', '▁a', 'co', 'ly', 'tes', '▁V', 'oodoo', '▁en', '▁Haïti', ',', '▁ou', '▁les', '▁gu', 'er', 're', 't', 's', '▁du', '▁désert', '▁K', 'ais', 'ut', 'ut', '▁du', '▁Nord', '▁du', '▁Kenya', ',', '▁le', '▁C', 'ura', 'nder', 'o', '▁dans', '▁les', '▁montagnes', '▁des', '▁And', 'es', ',', '▁ou', '▁une', '▁voiture', 'ava', 'n', 's', 'er', 'ai', '▁au', '▁milieu', '▁du', '▁Sahara', '▁--', '▁c', "'", 'est', '▁à', '▁peu', '▁près', '▁le', '▁gars', '▁que', '▁j', "'", 'ai', '▁voyagé', '▁dans', '▁le', '▁désert', '▁avec', '▁un', '▁mois', '▁--', '▁ou', '▁en', '▁effet', '▁un', '▁y', 'ak', '▁dans', '▁les', '▁pente', 's', '▁de', '▁Q', 'o', 'mo', 'lang', 'ma', ',', '▁de', '▁l', "'", 'Everest', ',', '▁la', '▁mère', '▁de', '▁la', '▁G', 'é', 't', 'in', 'ité', '▁du', '▁monde', '.'] 2024-01-16 12:38:47,862 - INFO - joeynmt.training - Example #2 2024-01-16 12:38:47,862 - INFO - joeynmt.training - Source: Aucun biologiste, par exemple, n'oserait suggérer que 50% ou plus de toutes les espèces ont été ou sont à deux doigts de l'extinction parce que ce n'est tout simplement pas vrai, et pourtant -- que le scénario le plus apocalyptique dans le royaume de la diversité biologique -- se rapproche rarement de ce que nous considérons comme le scénario le plus optimiste au sein de la diversité culturelle. 2024-01-16 12:38:47,863 - INFO - joeynmt.training - Reference: No biologists, for example, would dare suggest that 50 percent of all species or more have been or are on the brink of extinction because it simply is not true, and yet that -- the most apocalyptic scenario in the realm of biological diversity -- scarcely approaches what we know to be the most optimistic scenario in the realm of cultural diversity. 2024-01-16 12:38:47,863 - INFO - joeynmt.training - Hypothesis: No biologist, for example, would suggest that 50 or more of all species have been or two fingers of extinction because it's not just true, and yet -- that the most apocalyptic scenario in the realm of biological diversity -- rarely come to the point we think of as the most optimistic scenario in the breast of cultural diversity. 2024-01-16 12:38:47,863 - DEBUG - joeynmt.training - Tokenized source: ['▁Aucun', '▁biologist', 'e', ',', '▁par', '▁exemple', ',', '▁n', "'", 'ose', 'rait', '▁suggérer', '▁que', '▁50%', '▁ou', '▁plus', '▁de', '▁toutes', '▁les', '▁espèces', '▁ont', '▁été', '▁ou', '▁sont', '▁à', '▁deux', '▁doigts', '▁de', '▁l', "'", 'extinction', '▁parce', '▁que', '▁ce', '▁n', "'", 'est', '▁tout', '▁simplement', '▁pas', '▁vrai', ',', '▁et', '▁pourtant', '▁--', '▁que', '▁le', '▁scénario', '▁le', '▁plus', '▁a', 'po', 'ca', 'ly', 'p', 'tique', '▁dans', '▁le', '▁royaume', '▁de', '▁la', '▁diversité', '▁biologique', '▁--', '▁se', '▁rapproche', '▁rarement', '▁de', '▁ce', '▁que', '▁nous', '▁considér', 'ons', '▁comme', '▁le', '▁scénario', '▁le', '▁plus', '▁optimiste', '▁au', '▁sein', '▁de', '▁la', '▁diversité', '▁culturelle', '.'] 2024-01-16 12:38:47,864 - DEBUG - joeynmt.training - Tokenized reference: ['▁No', '▁biologists', ',', '▁for', '▁example', ',', '▁would', '▁da', 're', '▁suggest', '▁that', '▁50', '▁percent', '▁of', '▁all', '▁species', '▁or', '▁more', '▁have', '▁been', '▁or', '▁are', '▁on', '▁the', '▁brin', 'k', '▁of', '▁extinction', '▁because', '▁it', '▁simply', '▁is', '▁not', '▁true', ',', '▁and', '▁yet', '▁that', '▁--', '▁the', '▁most', '▁apocalyptic', '▁scenario', '▁in', '▁the', '▁realm', '▁of', '▁biological', '▁diversity', '▁--', '▁scarce', 'ly', '▁approaches', '▁what', '▁we', '▁know', '▁to', '▁be', '▁the', '▁most', '▁optimistic', '▁scenario', '▁in', '▁the', '▁realm', '▁of', '▁cultural', '▁diversity', '.'] 2024-01-16 12:38:47,864 - DEBUG - joeynmt.training - Tokenized hypothesis: ['', '▁And', '▁just', '▁as', '▁the', '▁biosphere', '▁has', '▁been', '▁severely', '▁er', 'od', 'ed', ',', '▁so', '▁too', '▁is', '▁the', '▁et', 'h', 'n', 'osphere', '▁--', '▁and', ',', '▁if', '▁anything', ',', '▁at', '▁a', '▁far', '▁greater', '▁rate', '.', '', '▁No', '▁biologist', ',', '▁for', '▁example', ',', '▁would', '▁suggest', '▁that', '▁50', '▁or', '▁more', '▁of', '▁all', '▁species', '▁have', '▁been', '▁or', '▁two', '▁fingers', '▁of', '▁extinction', '▁because', '▁it', "'", 's', '▁not', '▁just', '▁true', ',', '▁and', '▁yet', '▁--', '▁that', '▁the', '▁most', '▁apocalyptic', '▁scenario', '▁in', '▁the', '▁realm', '▁of', '▁biological', '▁diversity', '▁--', '▁rarely', '▁come', '▁to', '▁the', '▁point', '▁we', '▁think', '▁of', '▁as', '▁the', '▁most', '▁optimistic', '▁scenario', '▁in', '▁the', '▁breast', '▁of', '▁cultural', '▁diversity', '.'] 2024-01-16 12:38:47,864 - INFO - joeynmt.training - Example #3 2024-01-16 12:38:47,864 - INFO - joeynmt.training - Source: Et l'indicateur le plus fiable est bien sûr l'extinction du langage. 2024-01-16 12:38:47,864 - INFO - joeynmt.training - Reference: And the great indicator of that, of course, is language loss. 2024-01-16 12:38:47,864 - INFO - joeynmt.training - Hypothesis: And the most reliable indicator of this is obviously the extinction of language. 2024-01-16 12:38:47,865 - DEBUG - joeynmt.training - Tokenized source: ['▁Et', '▁l', "'", 'ind', 'ic', 'ateur', '▁le', '▁plus', '▁fiable', '▁est', '▁bien', '▁sûr', '▁l', "'", 'extinction', '▁du', '▁langage', '.'] 2024-01-16 12:38:47,865 - DEBUG - joeynmt.training - Tokenized reference: ['▁And', '▁the', '▁great', '▁indicator', '▁of', '▁that', ',', '▁of', '▁course', ',', '▁is', '▁language', '▁loss', '.'] 2024-01-16 12:38:47,865 - DEBUG - joeynmt.training - Tokenized hypothesis: ['', '▁No', '▁biologists', ',', '▁for', '▁example', ',', '▁would', '▁da', 're', '▁suggest', '▁that', '▁50', '▁percent', '▁of', '▁all', '▁species', '▁or', '▁more', '▁have', '▁been', '▁or', '▁are', '▁on', '▁the', '▁brin', 'k', '▁of', '▁extinction', '▁because', '▁it', '▁simply', '▁is', '▁not', '▁true', ',', '▁and', '▁yet', '▁that', '▁--', '▁the', '▁most', '▁apocalyptic', '▁scenario', '▁in', '▁the', '▁realm', '▁of', '▁biological', '▁diversity', '▁--', '▁scarce', 'ly', '▁approaches', '▁what', '▁we', '▁know', '▁to', '▁be', '▁the', '▁most', '▁optimistic', '▁scenario', '▁in', '▁the', '▁realm', '▁of', '▁cultural', '▁diversity', '.', '', '▁And', '▁the', '▁most', '▁reliable', '▁indicator', '▁of', '▁this', '▁is', '▁obviously', '▁the', '▁extinction', '▁of', '▁language', '.'] 2024-01-16 12:40:04,047 - INFO - joeynmt.training - Epoch 5, Step: 12100, Batch Loss: 1.470943, Batch Acc: 0.659314, Tokens per Sec: 16298, Lr: 0.000182 2024-01-16 12:41:20,100 - INFO - joeynmt.training - Epoch 5, Step: 12200, Batch Loss: 1.483927, Batch Acc: 0.658637, Tokens per Sec: 16376, Lr: 0.000181 2024-01-16 12:42:36,971 - INFO - joeynmt.training - Epoch 5, Step: 12300, Batch Loss: 1.519643, Batch Acc: 0.659948, Tokens per Sec: 16316, Lr: 0.000180 2024-01-16 12:43:53,572 - INFO - joeynmt.training - Epoch 5, Step: 12400, Batch Loss: 1.453635, Batch Acc: 0.659008, Tokens per Sec: 16314, Lr: 0.000180 2024-01-16 12:45:10,863 - INFO - joeynmt.training - Epoch 5, Step: 12500, Batch Loss: 1.457524, Batch Acc: 0.659738, Tokens per Sec: 16148, Lr: 0.000179 2024-01-16 12:46:28,543 - INFO - joeynmt.training - Epoch 5, Step: 12600, Batch Loss: 1.494839, Batch Acc: 0.661629, Tokens per Sec: 16160, Lr: 0.000178 2024-01-16 12:47:45,616 - INFO - joeynmt.training - Epoch 5, Step: 12700, Batch Loss: 1.462310, Batch Acc: 0.662100, Tokens per Sec: 16209, Lr: 0.000177 2024-01-16 12:49:03,575 - INFO - joeynmt.training - Epoch 5, Step: 12800, Batch Loss: 1.401735, Batch Acc: 0.664076, Tokens per Sec: 16091, Lr: 0.000177 2024-01-16 12:50:22,155 - INFO - joeynmt.training - Epoch 5, Step: 12900, Batch Loss: 1.446556, Batch Acc: 0.663910, Tokens per Sec: 15898, Lr: 0.000176 2024-01-16 12:51:39,807 - INFO - joeynmt.training - Epoch 5, Step: 13000, Batch Loss: 1.473133, Batch Acc: 0.664225, Tokens per Sec: 16177, Lr: 0.000175 2024-01-16 12:51:39,829 - INFO - joeynmt.datasets - Sample random subset from dev data: n=500, seed=13042 2024-01-16 12:51:39,838 - INFO - joeynmt.prediction - Predicting 500 example(s)... (Greedy decoding with min_output_length=1, max_output_length=512, return_prob='none', generate_unk=True, repetition_penalty=-1, no_repeat_ngram_size=-1) 2024-01-16 12:51:53,840 - INFO - joeynmt.prediction - Generation took 13.9905[sec]. 2024-01-16 12:51:54,361 - INFO - joeynmt.metrics - nrefs:1|case:lc|eff:no|tok:13a|smooth:exp|version:2.4.0 2024-01-16 12:51:54,362 - INFO - joeynmt.prediction - Evaluation result (greedy): bleu: 25.81, loss: 2.12, ppl: 8.32, acc: 0.57, 0.0798[sec] 2024-01-16 12:51:57,416 - INFO - joeynmt.training - Checkpoint saved in models/iwslt14_prompt/13000.ckpt. 2024-01-16 12:51:57,864 - INFO - joeynmt.training - Example #0 2024-01-16 12:51:57,865 - INFO - joeynmt.training - Source: And of course, we all share the same adaptive imperatives. 2024-01-16 12:51:57,865 - INFO - joeynmt.training - Reference: Bien sûr, nous partageons tous les mêmes impératifs d'adaptation. 2024-01-16 12:51:57,865 - INFO - joeynmt.training - Hypothesis: Et bien sûr, nous partageons tous les mêmes impératifs adaptatifs. 2024-01-16 12:51:57,866 - DEBUG - joeynmt.training - Tokenized source: ['▁And', '▁of', '▁course', ',', '▁we', '▁all', '▁share', '▁the', '▁same', '▁adaptive', '▁imperative', 's', '.'] 2024-01-16 12:51:57,866 - DEBUG - joeynmt.training - Tokenized reference: ['▁Bien', '▁sûr', ',', '▁nous', '▁partageons', '▁tous', '▁les', '▁mêmes', '▁im', 'pé', 'r', 'atif', 's', '▁d', "'", 'adaptation', '.'] 2024-01-16 12:51:57,872 - DEBUG - joeynmt.training - Tokenized hypothesis: ['', '▁Le', '▁fait', '▁de', '▁savoir', '▁que', '▁les', '▁Ja', 'gu', 'ar', '▁shaman', '▁voyage', 'nt', '▁toujours', '▁au', '-', 'delà', '▁de', '▁la', '▁voie', '▁l', 'act', 'ée', ',', '▁ou', '▁que', '▁les', '▁mythe', 's', '▁des', '▁anciens', '▁Inuit', '▁résonne', 'nt', '▁encore', '▁de', '▁sens', ',', '▁ou', '▁bien', '▁que', '▁dans', '▁l', "'", 'Himalaya', ',', '▁les', '▁Bo', 'ud', 'd', 'h', 'istes', '▁continuent', '▁à', '▁recherche', 'r', '▁le', '▁souffle', '▁du', '▁D', 'ha', 'r', 'ma', ',', '▁c', "'", 'est', '▁se', '▁rappeler', '▁de', '▁la', '▁révélation', '▁essentielle', '▁de', '▁l', "'", 'anthrop', 'ologie', ',', '▁et', '▁cela', '▁veut', '▁dire', '▁que', '▁le', '▁monde', '▁dans', '▁lequel', '▁nous', '▁vivons', '▁n', "'", 'existe', '▁pas', '▁dans', '▁un', '▁sens', '▁absolu', ',', '▁mais', '▁est', '▁uniquement', '▁un', '▁exemple', '▁de', '▁réalité', ',', '▁la', '▁conséquence', '▁d', "'", 'un', '▁ensemble', '▁spécifique', '▁de', '▁choix', '▁adapté', 's', '▁établi', 's', '▁par', '▁notre', '▁lignée', '▁avec', '▁succès', ',', '▁il', '▁y', '▁a', '▁plusieurs', '▁générations', '.', '', '▁Et', '▁bien', '▁sûr', ',', '▁nous', '▁partageons', '▁tous', '▁les', '▁mêmes', '▁im', 'pé', 'r', 'atif', 's', '▁adapt', 'atif', 's', '.'] 2024-01-16 12:51:57,872 - INFO - joeynmt.training - Example #1 2024-01-16 12:51:57,872 - INFO - joeynmt.training - Source: And this is an idea, if you think about it, can only fill you with hope. 2024-01-16 12:51:57,872 - INFO - joeynmt.training - Reference: Et ceci est une idée, si on y réfléchit, qui ne peut que vous remplir d'espoir. 2024-01-16 12:51:57,872 - INFO - joeynmt.training - Hypothesis: Et c'est une idée, si vous y réfléchissez, ne peut vous remplir que de l'espoir. 2024-01-16 12:51:57,873 - DEBUG - joeynmt.training - Tokenized source: ['▁And', '▁this', '▁is', '▁an', '▁idea', ',', '▁if', '▁you', '▁think', '▁about', '▁it', ',', '▁can', '▁only', '▁fill', '▁you', '▁with', '▁hope', '.'] 2024-01-16 12:51:57,873 - DEBUG - joeynmt.training - Tokenized reference: ['▁Et', '▁ceci', '▁est', '▁une', '▁idée', ',', '▁si', '▁on', '▁y', '▁réfléchi', 't', ',', '▁qui', '▁ne', '▁peut', '▁que', '▁vous', '▁remplir', '▁d', "'", 'espoir', '.'] 2024-01-16 12:51:57,873 - DEBUG - joeynmt.training - Tokenized hypothesis: ['', '▁Tous', '▁ces', '▁peuple', 's', '▁nous', '▁enseigne', 'nt', '▁qu', "'", 'il', '▁y', '▁a', '▁d', "'", 'autres', '▁façons', '▁d', "'", 'être', ',', '▁d', "'", 'autres', '▁façons', '▁de', '▁penser', ',', '▁d', "'", 'autres', '▁manières', '▁de', '▁s', "'", 'orient', 'er', '▁sur', '▁Terre', '.', '', '▁Et', '▁c', "'", 'est', '▁une', '▁idée', ',', '▁si', '▁vous', '▁y', '▁réfléchissez', ',', '▁ne', '▁peut', '▁vous', '▁remplir', '▁que', '▁de', '▁l', "'", 'espoir', '.'] 2024-01-16 12:51:57,873 - INFO - joeynmt.training - Example #2 2024-01-16 12:51:57,874 - INFO - joeynmt.training - Source: And yet, that dreadful fate is indeed the plight of somebody somewhere on Earth roughly every two weeks, because every two weeks, some elder dies and carries with him into the grave the last syllables of an ancient tongue. 2024-01-16 12:51:57,874 - INFO - joeynmt.training - Reference: Et pourtant, cette atroce fatalité est en effet le désespoir de quelqu'un quelque part sur terre, tous les quinze jours à peu près, parce que tous les quinze jours, un ancien meurt et emporte les dernières syllabes avec lui dans la tombe d'une langue ancienne. 2024-01-16 12:51:57,874 - INFO - joeynmt.training - Hypothesis: Et pourtant, ce sort de voile est en effet le pâteau de quelqu'un quelque part sur Terre en gros toutes les deux semaines, parce que tous les deux semaines, certains anciens meurent et se portent avec lui dans la tombe la dernière syllable de la langue ancienne. 2024-01-16 12:51:57,875 - DEBUG - joeynmt.training - Tokenized source: ['▁And', '▁yet', ',', '▁that', '▁dreadful', '▁fate', '▁is', '▁indeed', '▁the', '▁p', 'light', '▁of', '▁somebody', '▁somewhere', '▁on', '▁Earth', '▁roughly', '▁every', '▁two', '▁weeks', ',', '▁because', '▁every', '▁two', '▁weeks', ',', '▁some', '▁el', 'der', '▁die', 's', '▁and', '▁carries', '▁with', '▁him', '▁into', '▁the', '▁grave', '▁the', '▁last', '▁s', 'y', 'll', 'ables', '▁of', '▁an', '▁ancient', '▁tongue', '.'] 2024-01-16 12:51:57,875 - DEBUG - joeynmt.training - Tokenized reference: ['▁Et', '▁pourtant', ',', '▁cette', '▁atroce', '▁fatal', 'ité', '▁est', '▁en', '▁effet', '▁le', '▁désespoir', '▁de', '▁quelqu', "'", 'un', '▁quelque', '▁part', '▁sur', '▁terre', ',', '▁tous', '▁les', '▁qui', 'n', 'ze', '▁jours', '▁à', '▁peu', '▁près', ',', '▁parce', '▁que', '▁tous', '▁les', '▁qui', 'n', 'ze', '▁jours', ',', '▁un', '▁ancien', '▁meurt', '▁et', '▁', 'emporte', '▁les', '▁dernières', '▁s', 'y', 'lla', 'be', 's', '▁avec', '▁lui', '▁dans', '▁la', '▁tombe', '▁d', "'", 'une', '▁langue', '▁ancienne', '.'] 2024-01-16 12:51:57,875 - DEBUG - joeynmt.training - Tokenized hypothesis: ['', '▁Comment', '▁ne', '▁pas', '▁se', '▁sentir', '▁seuls', ',', '▁en', 've', 'lop', 'pé', 's', '▁dans', '▁le', '▁silence', ',', '▁et', '▁être', '▁le', '▁dernier', '▁de', '▁votre', '▁peuple', '▁à', '▁parler', '▁votre', '▁langue', ',', '▁de', '▁n', "'", 'avoir', '▁aucun', '▁moyen', '▁de', '▁transmettre', '▁la', '▁sagesse', '▁des', '▁ancêtres', '▁ou', '▁de', 'van', 'c', 'er', '▁la', '▁promesse', '▁des', '▁enfants', '▁?', '', '▁Et', '▁pourtant', ',', '▁ce', '▁sort', '▁de', '▁voile', '▁est', '▁en', '▁effet', '▁le', '▁p', 'ât', 'eau', '▁de', '▁quelqu', "'", 'un', '▁quelque', '▁part', '▁sur', '▁Terre', '▁en', '▁gros', '▁toutes', '▁les', '▁deux', '▁semaines', ',', '▁parce', '▁que', '▁tous', '▁les', '▁deux', '▁semaines', ',', '▁certains', '▁anciens', '▁meurent', '▁et', '▁se', '▁portent', '▁avec', '▁lui', '▁dans', '▁la', '▁tombe', '▁la', '▁dernière', '▁s', 'y', 'll', 'able', '▁de', '▁la', '▁langue', '▁ancienne', '.'] 2024-01-16 12:51:57,875 - INFO - joeynmt.training - Example #3 2024-01-16 12:51:57,875 - INFO - joeynmt.training - Source: Du Kogi." 2024-01-16 12:51:57,875 - INFO - joeynmt.training - Reference: Let's make it Kogi." 2024-01-16 12:51:57,875 - INFO - joeynmt.training - Hypothesis: Du Kogi." 2024-01-16 12:51:57,876 - DEBUG - joeynmt.training - Tokenized source: ['▁Du', '▁Ko', 'g', 'i', '."'] 2024-01-16 12:51:57,876 - DEBUG - joeynmt.training - Tokenized reference: ['▁Let', "'", 's', '▁make', '▁it', '▁Ko', 'g', 'i', '."'] 2024-01-16 12:51:57,876 - DEBUG - joeynmt.training - Tokenized hypothesis: ['', '▁And', '▁I', '▁know', '▁there', "'", 's', '▁some', '▁of', '▁you', '▁who', '▁say', ',', '▁"', 'Well', ',', '▁would', 'n', "'", 't', '▁it', '▁be', '▁better', ',', '▁would', 'n', "'", 't', '▁the', '▁world', '▁be', '▁a', '▁better', '▁place', '▁if', '▁we', '▁all', '▁just', '▁spoke', '▁one', '▁language', '?"', '▁And', '▁I', '▁say', ',', '▁"', 'Great', ',', '▁let', "'", 's', '▁make', '▁that', '▁language', '▁Yoruba', '.', '▁Let', "'", 's', '▁make', '▁it', '▁Can', 'ton', 'es', 'e', '.', '', '▁Du', '▁Ko', 'g', 'i', '."'] 2024-01-16 12:53:14,801 - INFO - joeynmt.training - Epoch 5, Step: 13100, Batch Loss: 1.392111, Batch Acc: 0.665328, Tokens per Sec: 16281, Lr: 0.000175 2024-01-16 12:54:31,027 - INFO - joeynmt.training - Epoch 5, Step: 13200, Batch Loss: 1.464073, Batch Acc: 0.667028, Tokens per Sec: 16338, Lr: 0.000174 2024-01-16 12:55:47,309 - INFO - joeynmt.training - Epoch 5, Step: 13300, Batch Loss: 1.438363, Batch Acc: 0.667251, Tokens per Sec: 16377, Lr: 0.000173 2024-01-16 12:57:03,777 - INFO - joeynmt.training - Epoch 5, Step: 13400, Batch Loss: 1.470552, Batch Acc: 0.668765, Tokens per Sec: 16280, Lr: 0.000173 2024-01-16 12:58:19,829 - INFO - joeynmt.training - Epoch 5, Step: 13500, Batch Loss: 1.411500, Batch Acc: 0.667244, Tokens per Sec: 16433, Lr: 0.000172 2024-01-16 12:59:37,414 - INFO - joeynmt.training - Epoch 5, Step: 13600, Batch Loss: 1.457347, Batch Acc: 0.670312, Tokens per Sec: 16118, Lr: 0.000171 2024-01-16 13:00:57,621 - INFO - joeynmt.training - Epoch 5, Step: 13700, Batch Loss: 1.363226, Batch Acc: 0.671489, Tokens per Sec: 15628, Lr: 0.000171 2024-01-16 13:01:09,215 - INFO - joeynmt.training - Epoch 5, total training loss: 4051.75, num. of seqs: 702202, num. of tokens: 34266040, 2108.1706[sec] 2024-01-16 13:01:09,287 - INFO - joeynmt.training - EPOCH 6 2024-01-16 13:02:13,905 - INFO - joeynmt.training - Epoch 6, Step: 13800, Batch Loss: 1.334254, Batch Acc: 0.686187, Tokens per Sec: 16486, Lr: 0.000170 2024-01-16 13:03:30,951 - INFO - joeynmt.training - Epoch 6, Step: 13900, Batch Loss: 1.342006, Batch Acc: 0.683704, Tokens per Sec: 16211, Lr: 0.000170 2024-01-16 13:04:46,664 - INFO - joeynmt.training - Epoch 6, Step: 14000, Batch Loss: 1.403432, Batch Acc: 0.684877, Tokens per Sec: 16461, Lr: 0.000169 2024-01-16 13:04:46,684 - INFO - joeynmt.datasets - Sample random subset from dev data: n=500, seed=14042 2024-01-16 13:04:46,692 - INFO - joeynmt.prediction - Predicting 500 example(s)... (Greedy decoding with min_output_length=1, max_output_length=512, return_prob='none', generate_unk=True, repetition_penalty=-1, no_repeat_ngram_size=-1) 2024-01-16 13:05:04,604 - INFO - joeynmt.prediction - Generation took 17.9107[sec]. 2024-01-16 13:05:04,839 - INFO - joeynmt.metrics - nrefs:1|case:lc|eff:no|tok:13a|smooth:exp|version:2.4.0 2024-01-16 13:05:04,840 - INFO - joeynmt.prediction - Evaluation result (greedy): bleu: 27.53, loss: 2.13, ppl: 8.44, acc: 0.57, 0.1146[sec] 2024-01-16 13:05:04,840 - INFO - joeynmt.training - Hooray! New best validation result [bleu]! 2024-01-16 13:05:08,049 - INFO - joeynmt.training - Checkpoint saved in models/iwslt14_prompt/14000.ckpt. 2024-01-16 13:05:08,053 - INFO - joeynmt.training - Example #0 2024-01-16 13:05:08,054 - INFO - joeynmt.training - Source: And of course, we all share the same adaptive imperatives. 2024-01-16 13:05:08,054 - INFO - joeynmt.training - Reference: Bien sûr, nous partageons tous les mêmes impératifs d'adaptation. 2024-01-16 13:05:08,054 - INFO - joeynmt.training - Hypothesis: Et bien sûr, nous partageons tous les mêmes impératifs adaptatifs. 2024-01-16 13:05:08,055 - DEBUG - joeynmt.training - Tokenized source: ['▁And', '▁of', '▁course', ',', '▁we', '▁all', '▁share', '▁the', '▁same', '▁adaptive', '▁imperative', 's', '.'] 2024-01-16 13:05:08,055 - DEBUG - joeynmt.training - Tokenized reference: ['▁Bien', '▁sûr', ',', '▁nous', '▁partageons', '▁tous', '▁les', '▁mêmes', '▁im', 'pé', 'r', 'atif', 's', '▁d', "'", 'adaptation', '.'] 2024-01-16 13:05:08,055 - DEBUG - joeynmt.training - Tokenized hypothesis: ['', '▁Le', '▁fait', '▁de', '▁savoir', '▁que', '▁les', '▁Ja', 'gu', 'ar', '▁shaman', '▁voyage', 'nt', '▁toujours', '▁au', '-', 'delà', '▁de', '▁la', '▁voie', '▁l', 'act', 'ée', ',', '▁ou', '▁que', '▁les', '▁mythe', 's', '▁des', '▁anciens', '▁Inuit', '▁résonne', 'nt', '▁encore', '▁de', '▁sens', ',', '▁ou', '▁bien', '▁que', '▁dans', '▁l', "'", 'Himalaya', ',', '▁les', '▁Bo', 'ud', 'd', 'h', 'istes', '▁continuent', '▁à', '▁recherche', 'r', '▁le', '▁souffle', '▁du', '▁D', 'ha', 'r', 'ma', ',', '▁c', "'", 'est', '▁se', '▁rappeler', '▁de', '▁la', '▁révélation', '▁essentielle', '▁de', '▁l', "'", 'anthrop', 'ologie', ',', '▁et', '▁cela', '▁veut', '▁dire', '▁que', '▁le', '▁monde', '▁dans', '▁lequel', '▁nous', '▁vivons', '▁n', "'", 'existe', '▁pas', '▁dans', '▁un', '▁sens', '▁absolu', ',', '▁mais', '▁est', '▁uniquement', '▁un', '▁exemple', '▁de', '▁réalité', ',', '▁la', '▁conséquence', '▁d', "'", 'un', '▁ensemble', '▁spécifique', '▁de', '▁choix', '▁adapté', 's', '▁établi', 's', '▁par', '▁notre', '▁lignée', '▁avec', '▁succès', ',', '▁il', '▁y', '▁a', '▁plusieurs', '▁générations', '.', '', '▁Et', '▁bien', '▁sûr', ',', '▁nous', '▁partageons', '▁tous', '▁les', '▁mêmes', '▁im', 'pé', 'r', 'atif', 's', '▁adapt', 'atif', 's', '.'] 2024-01-16 13:05:08,055 - INFO - joeynmt.training - Example #1 2024-01-16 13:05:08,056 - INFO - joeynmt.training - Source: Du Kogi." 2024-01-16 13:05:08,056 - INFO - joeynmt.training - Reference: Let's make it Kogi." 2024-01-16 13:05:08,056 - INFO - joeynmt.training - Hypothesis: Du Kogi." 2024-01-16 13:05:08,057 - DEBUG - joeynmt.training - Tokenized source: ['▁Du', '▁Ko', 'g', 'i', '."'] 2024-01-16 13:05:08,057 - DEBUG - joeynmt.training - Tokenized reference: ['▁Let', "'", 's', '▁make', '▁it', '▁Ko', 'g', 'i', '."'] 2024-01-16 13:05:08,057 - DEBUG - joeynmt.training - Tokenized hypothesis: ['', '▁And', '▁I', '▁know', '▁there', "'", 's', '▁some', '▁of', '▁you', '▁who', '▁say', ',', '▁"', 'Well', ',', '▁would', 'n', "'", 't', '▁it', '▁be', '▁better', ',', '▁would', 'n', "'", 't', '▁the', '▁world', '▁be', '▁a', '▁better', '▁place', '▁if', '▁we', '▁all', '▁just', '▁spoke', '▁one', '▁language', '?"', '▁And', '▁I', '▁say', ',', '▁"', 'Great', ',', '▁let', "'", 's', '▁make', '▁that', '▁language', '▁Yoruba', '.', '▁Let', "'", 's', '▁make', '▁it', '▁Can', 'ton', 'es', 'e', '.', '', '▁Du', '▁Ko', 'g', 'i', '."'] 2024-01-16 13:05:08,057 - INFO - joeynmt.training - Example #2 2024-01-16 13:05:08,058 - INFO - joeynmt.training - Source: Beaucoup d'entre nous oublient un peu que lorsque je dis "des façons différentes d'être", je veux vraiment dire des façons différentes d'être. 2024-01-16 13:05:08,058 - INFO - joeynmt.training - Reference: Now, there are many of us who sort of forget that when I say "different ways of being," I really do mean different ways of being. 2024-01-16 13:05:08,058 - INFO - joeynmt.training - Hypothesis: Many of us forget a little bit about when I say, "a different way of being," I really want to say different ways of being. 2024-01-16 13:05:08,059 - DEBUG - joeynmt.training - Tokenized source: ['▁Beaucoup', '▁d', "'", 'entre', '▁nous', '▁', 'oubli', 'ent', '▁un', '▁peu', '▁que', '▁lorsque', '▁je', '▁dis', '▁"', 'des', '▁façons', '▁différentes', '▁d', "'", 'être', '",', '▁je', '▁veux', '▁vraiment', '▁dire', '▁des', '▁façons', '▁différentes', '▁d', "'", 'être', '.'] 2024-01-16 13:05:08,059 - DEBUG - joeynmt.training - Tokenized reference: ['▁Now', ',', '▁there', '▁are', '▁many', '▁of', '▁us', '▁who', '▁sort', '▁of', '▁forget', '▁that', '▁when', '▁I', '▁say', '▁"', 'd', 'iff', 'er', 'ent', '▁ways', '▁of', '▁being', ',"', '▁I', '▁really', '▁do', '▁mean', '▁different', '▁ways', '▁of', '▁being', '.'] 2024-01-16 13:05:08,059 - DEBUG - joeynmt.training - Tokenized hypothesis: ['', '▁And', '▁so', ',', '▁what', '▁I', "'", 'd', '▁like', '▁to', '▁do', '▁with', '▁you', '▁today', '▁is', '▁sort', '▁of', '▁take', '▁you', '▁on', '▁a', '▁journey', '▁through', '▁the', '▁et', 'h', 'n', 'osphere', ',', '▁a', '▁brief', '▁journey', '▁through', '▁the', '▁et', 'h', 'n', 'osphere', ',', '▁to', '▁try', '▁to', '▁begin', '▁to', '▁give', '▁you', '▁a', '▁sense', '▁of', '▁what', '▁in', '▁fact', '▁is', '▁being', '▁lost', '.', '', '▁Many', '▁of', '▁us', '▁forget', '▁a', '▁little', '▁bit', '▁about', '▁when', '▁I', '▁say', ',', '▁"', 'a', '▁different', '▁way', '▁of', '▁being', ',"', '▁I', '▁really', '▁want', '▁to', '▁say', '▁different', '▁ways', '▁of', '▁being', '.'] 2024-01-16 13:05:08,059 - INFO - joeynmt.training - Example #3 2024-01-16 13:05:08,060 - INFO - joeynmt.training - Source: They have a curious language and marriage rule which is called "linguistic exogamy:" you must marry someone who speaks a different language. 2024-01-16 13:05:08,060 - INFO - joeynmt.training - Reference: Ils ont une règle de langue et de mariage particulières qui s'appelle l'exogamie linguistique : vous devez épouser une personne qui parle une langue différente. 2024-01-16 13:05:08,060 - INFO - joeynmt.training - Hypothesis: Ils ont une politique curieuse et une règle du mariage qui s'appelle l'exogmie. Vous devez épouser quelqu'un qui parle une langue différente. 2024-01-16 13:05:08,060 - DEBUG - joeynmt.training - Tokenized source: ['▁They', '▁have', '▁a', '▁curious', '▁language', '▁and', '▁marriage', '▁rule', '▁which', '▁is', '▁called', '▁"', 'ling', 'u', 'istic', '▁ex', 'o', 'ga', 'm', 'y', ':', '"', '▁you', '▁must', '▁marry', '▁someone', '▁who', '▁speak', 's', '▁a', '▁different', '▁language', '.'] 2024-01-16 13:05:08,061 - DEBUG - joeynmt.training - Tokenized reference: ['▁Ils', '▁ont', '▁une', '▁règle', '▁de', '▁langue', '▁et', '▁de', '▁mariage', '▁particulière', 's', '▁qui', '▁s', "'", 'appelle', '▁l', "'", 'ex', 'o', 'g', 'ami', 'e', '▁linguistique', '▁:', '▁vous', '▁devez', '▁épouse', 'r', '▁une', '▁personne', '▁qui', '▁parle', '▁une', '▁langue', '▁différente', '.'] 2024-01-16 13:05:08,061 - DEBUG - joeynmt.training - Tokenized hypothesis: ['', '▁C', "'", 'est', '▁un', '▁peuple', '▁dont', '▁l', "'", 'état', '▁des', '▁connaissances', '▁ne', '▁permet', '▁pas', '▁de', '▁faire', '▁la', '▁distinction', '▁entre', '▁la', '▁couleur', '▁bleue', '▁et', '▁verte', '▁parce', '▁que', '▁la', '▁canopée', '▁des', '▁c', 'ieux', '▁est', '▁égale', '▁à', '▁la', '▁canopée', '▁de', '▁la', '▁forêt', '▁dont', '▁le', '▁peuple', '▁dépend', '.', '', '▁Ils', '▁ont', '▁une', '▁politique', '▁curieuse', '▁et', '▁une', '▁règle', '▁du', '▁mariage', '▁qui', '▁s', "'", 'appelle', '▁l', "'", 'ex', 'o', 'g', 'm', 'ie', '.', '▁Vous', '▁devez', '▁épouse', 'r', '▁quelqu', "'", 'un', '▁qui', '▁parle', '▁une', '▁langue', '▁différente', '.'] 2024-01-16 13:06:26,031 - INFO - joeynmt.training - Epoch 6, Step: 14100, Batch Loss: 1.305937, Batch Acc: 0.683632, Tokens per Sec: 16139, Lr: 0.000168 2024-01-16 13:07:43,100 - INFO - joeynmt.training - Epoch 6, Step: 14200, Batch Loss: 1.412088, Batch Acc: 0.684462, Tokens per Sec: 16208, Lr: 0.000168 2024-01-16 13:08:59,742 - INFO - joeynmt.training - Epoch 6, Step: 14300, Batch Loss: 1.395634, Batch Acc: 0.684173, Tokens per Sec: 16299, Lr: 0.000167 2024-01-16 13:10:16,339 - INFO - joeynmt.training - Epoch 6, Step: 14400, Batch Loss: 1.343596, Batch Acc: 0.683810, Tokens per Sec: 16233, Lr: 0.000167 2024-01-16 13:11:33,719 - INFO - joeynmt.training - Epoch 6, Step: 14500, Batch Loss: 1.294144, Batch Acc: 0.684436, Tokens per Sec: 16162, Lr: 0.000166 2024-01-16 13:12:50,272 - INFO - joeynmt.training - Epoch 6, Step: 14600, Batch Loss: 1.389207, Batch Acc: 0.685961, Tokens per Sec: 16327, Lr: 0.000166 2024-01-16 13:14:07,331 - INFO - joeynmt.training - Epoch 6, Step: 14700, Batch Loss: 1.340038, Batch Acc: 0.684664, Tokens per Sec: 16260, Lr: 0.000165 2024-01-16 13:15:23,643 - INFO - joeynmt.training - Epoch 6, Step: 14800, Batch Loss: 1.323851, Batch Acc: 0.686472, Tokens per Sec: 16298, Lr: 0.000164 2024-01-16 13:16:40,250 - INFO - joeynmt.training - Epoch 6, Step: 14900, Batch Loss: 1.325114, Batch Acc: 0.685745, Tokens per Sec: 16231, Lr: 0.000164 2024-01-16 13:17:55,637 - INFO - joeynmt.training - Epoch 6, Step: 15000, Batch Loss: 1.320591, Batch Acc: 0.687525, Tokens per Sec: 16438, Lr: 0.000163 2024-01-16 13:17:55,638 - INFO - joeynmt.datasets - Sample random subset from dev data: n=500, seed=15042 2024-01-16 13:17:55,638 - INFO - joeynmt.prediction - Predicting 500 example(s)... (Greedy decoding with min_output_length=1, max_output_length=512, return_prob='none', generate_unk=True, repetition_penalty=-1, no_repeat_ngram_size=-1) 2024-01-16 13:18:09,728 - INFO - joeynmt.prediction - Generation took 14.0812[sec]. 2024-01-16 13:18:10,963 - INFO - joeynmt.metrics - nrefs:1|case:lc|eff:no|tok:13a|smooth:exp|version:2.4.0 2024-01-16 13:18:10,963 - INFO - joeynmt.prediction - Evaluation result (greedy): bleu: 25.90, loss: 2.10, ppl: 8.18, acc: 0.57, 1.2183[sec] 2024-01-16 13:18:13,889 - INFO - joeynmt.training - Checkpoint saved in models/iwslt14_prompt/15000.ckpt. 2024-01-16 13:18:14,288 - INFO - joeynmt.training - Example #0 2024-01-16 13:18:14,289 - INFO - joeynmt.training - Source: But what's interesting is the unique cadence of the song, the rhythm of the dance in every culture. 2024-01-16 13:18:14,289 - INFO - joeynmt.training - Reference: Par contre, la cadence exceptionnelle de la chanson est intéressante, le rythme de la danse dans toutes les cultures. 2024-01-16 13:18:14,289 - INFO - joeynmt.training - Hypothesis: Mais ce qui est intéressant c'est la cadence unique de la chanson, le rythme de la danse dans toutes les cultures. 2024-01-16 13:18:14,291 - DEBUG - joeynmt.training - Tokenized source: ['▁But', '▁what', "'", 's', '▁interesting', '▁is', '▁the', '▁unique', '▁cadence', '▁of', '▁the', '▁song', ',', '▁the', '▁rhythm', '▁of', '▁the', '▁dance', '▁in', '▁every', '▁culture', '.'] 2024-01-16 13:18:14,291 - DEBUG - joeynmt.training - Tokenized reference: ['▁Par', '▁contre', ',', '▁la', '▁cadence', '▁exceptionnel', 'le', '▁de', '▁la', '▁chanson', '▁est', '▁intéressante', ',', '▁le', '▁rythme', '▁de', '▁la', '▁dans', 'e', '▁dans', '▁toutes', '▁les', '▁cultures', '.'] 2024-01-16 13:18:14,291 - DEBUG - joeynmt.training - Tokenized hypothesis: ['', '▁Nous', '▁devons', '▁faire', '▁face', '▁à', '▁la', '▁séparation', '▁inexorable', '▁de', '▁la', '▁mort', ',', '▁cela', '▁ne', '▁devrait', '▁donc', '▁pas', '▁nous', '▁surprend', 're', '▁que', '▁nous', '▁chant', 'ions', ',', '▁nous', '▁dans', 'ions', ',', '▁nous', '▁sommes', '▁tous', '▁des', '▁artistes', '.', '', '▁Mais', '▁ce', '▁qui', '▁est', '▁intéressant', '▁c', "'", 'est', '▁la', '▁cadence', '▁unique', '▁de', '▁la', '▁chanson', ',', '▁le', '▁rythme', '▁de', '▁la', '▁dans', 'e', '▁dans', '▁toutes', '▁les', '▁cultures', '.'] 2024-01-16 13:18:14,292 - INFO - joeynmt.training - Example #1 2024-01-16 13:18:14,292 - INFO - joeynmt.training - Source: No biologists, for example, would dare suggest that 50 percent of all species or more have been or are on the brink of extinction because it simply is not true, and yet that -- the most apocalyptic scenario in the realm of biological diversity -- scarcely approaches what we know to be the most optimistic scenario in the realm of cultural diversity. 2024-01-16 13:18:14,292 - INFO - joeynmt.training - Reference: Aucun biologiste, par exemple, n'oserait suggérer que 50% ou plus de toutes les espèces ont été ou sont à deux doigts de l'extinction parce que ce n'est tout simplement pas vrai, et pourtant -- que le scénario le plus apocalyptique dans le royaume de la diversité biologique -- se rapproche rarement de ce que nous considérons comme le scénario le plus optimiste au sein de la diversité culturelle. 2024-01-16 13:18:14,292 - INFO - joeynmt.training - Hypothesis: Les biologistes, par exemple, suggèrent que 50 % de toutes les espèces ou plus ont été ou sont sur le bord de l'extinction parce que c'est tout simplement vrai, et pourtant cela -- le scénario le plus aptasé dans le royaume de la diversité biologique -- à peine, approchent ce que nous savons être le scénario le plus optimiste dans le royaume de la diversité culturelle. 2024-01-16 13:18:14,293 - DEBUG - joeynmt.training - Tokenized source: ['▁No', '▁biologists', ',', '▁for', '▁example', ',', '▁would', '▁da', 're', '▁suggest', '▁that', '▁50', '▁percent', '▁of', '▁all', '▁species', '▁or', '▁more', '▁have', '▁been', '▁or', '▁are', '▁on', '▁the', '▁brin', 'k', '▁of', '▁extinction', '▁because', '▁it', '▁simply', '▁is', '▁not', '▁true', ',', '▁and', '▁yet', '▁that', '▁--', '▁the', '▁most', '▁apocalyptic', '▁scenario', '▁in', '▁the', '▁realm', '▁of', '▁biological', '▁diversity', '▁--', '▁scarce', 'ly', '▁approaches', '▁what', '▁we', '▁know', '▁to', '▁be', '▁the', '▁most', '▁optimistic', '▁scenario', '▁in', '▁the', '▁realm', '▁of', '▁cultural', '▁diversity', '.'] 2024-01-16 13:18:14,293 - DEBUG - joeynmt.training - Tokenized reference: ['▁Aucun', '▁biologist', 'e', ',', '▁par', '▁exemple', ',', '▁n', "'", 'ose', 'rait', '▁suggérer', '▁que', '▁50%', '▁ou', '▁plus', '▁de', '▁toutes', '▁les', '▁espèces', '▁ont', '▁été', '▁ou', '▁sont', '▁à', '▁deux', '▁doigts', '▁de', '▁l', "'", 'extinction', '▁parce', '▁que', '▁ce', '▁n', "'", 'est', '▁tout', '▁simplement', '▁pas', '▁vrai', ',', '▁et', '▁pourtant', '▁--', '▁que', '▁le', '▁scénario', '▁le', '▁plus', '▁a', 'po', 'ca', 'ly', 'p', 'tique', '▁dans', '▁le', '▁royaume', '▁de', '▁la', '▁diversité', '▁biologique', '▁--', '▁se', '▁rapproche', '▁rarement', '▁de', '▁ce', '▁que', '▁nous', '▁considér', 'ons', '▁comme', '▁le', '▁scénario', '▁le', '▁plus', '▁optimiste', '▁au', '▁sein', '▁de', '▁la', '▁diversité', '▁culturelle', '.'] 2024-01-16 13:18:14,293 - DEBUG - joeynmt.training - Tokenized hypothesis: ['', '▁Et', '▁lorsque', '▁la', '▁biosphère', '▁fut', '▁sérieusement', '▁é', 'ro', 'd', 'ée', ',', '▁l', "'", 'e', 'th', 'no', 's', 'ph', 'ère', '▁l', "'", 'a', '▁été', '▁également', '▁--', '▁et', '▁peut', '-', 'être', '▁bien', '▁plus', '▁rapidement', '.', '', '▁Les', '▁biologistes', ',', '▁par', '▁exemple', ',', '▁suggère', 'nt', '▁que', '▁50', '▁%', '▁de', '▁toutes', '▁les', '▁espèces', '▁ou', '▁plus', '▁ont', '▁été', '▁ou', '▁sont', '▁sur', '▁le', '▁bord', '▁de', '▁l', "'", 'extinction', '▁parce', '▁que', '▁c', "'", 'est', '▁tout', '▁simplement', '▁vrai', ',', '▁et', '▁pourtant', '▁cela', '▁--', '▁le', '▁scénario', '▁le', '▁plus', '▁a', 'p', 'ta', 's', 'é', '▁dans', '▁le', '▁royaume', '▁de', '▁la', '▁diversité', '▁biologique', '▁--', '▁à', '▁peine', ',', '▁approche', 'nt', '▁ce', '▁que', '▁nous', '▁savons', '▁être', '▁le', '▁scénario', '▁le', '▁plus', '▁optimiste', '▁dans', '▁le', '▁royaume', '▁de', '▁la', '▁diversité', '▁culturelle', '.'] 2024-01-16 13:18:14,293 - INFO - joeynmt.training - Example #2 2024-01-16 13:18:14,294 - INFO - joeynmt.training - Source: And yet, that dreadful fate is indeed the plight of somebody somewhere on Earth roughly every two weeks, because every two weeks, some elder dies and carries with him into the grave the last syllables of an ancient tongue. 2024-01-16 13:18:14,294 - INFO - joeynmt.training - Reference: Et pourtant, cette atroce fatalité est en effet le désespoir de quelqu'un quelque part sur terre, tous les quinze jours à peu près, parce que tous les quinze jours, un ancien meurt et emporte les dernières syllabes avec lui dans la tombe d'une langue ancienne. 2024-01-16 13:18:14,294 - INFO - joeynmt.training - Hypothesis: Et pourtant, ce destin horrible est en fait le plaisir de quelqu'un quelque part sur Terre, en gros tous les deux semaines, parce que tous les deux semaines, certains meurt et se porte avec lui dans la tombe le dernier syllable d'une langue ancienne. 2024-01-16 13:18:14,295 - DEBUG - joeynmt.training - Tokenized source: ['▁And', '▁yet', ',', '▁that', '▁dreadful', '▁fate', '▁is', '▁indeed', '▁the', '▁p', 'light', '▁of', '▁somebody', '▁somewhere', '▁on', '▁Earth', '▁roughly', '▁every', '▁two', '▁weeks', ',', '▁because', '▁every', '▁two', '▁weeks', ',', '▁some', '▁el', 'der', '▁die', 's', '▁and', '▁carries', '▁with', '▁him', '▁into', '▁the', '▁grave', '▁the', '▁last', '▁s', 'y', 'll', 'ables', '▁of', '▁an', '▁ancient', '▁tongue', '.'] 2024-01-16 13:18:14,295 - DEBUG - joeynmt.training - Tokenized reference: ['▁Et', '▁pourtant', ',', '▁cette', '▁atroce', '▁fatal', 'ité', '▁est', '▁en', '▁effet', '▁le', '▁désespoir', '▁de', '▁quelqu', "'", 'un', '▁quelque', '▁part', '▁sur', '▁terre', ',', '▁tous', '▁les', '▁qui', 'n', 'ze', '▁jours', '▁à', '▁peu', '▁près', ',', '▁parce', '▁que', '▁tous', '▁les', '▁qui', 'n', 'ze', '▁jours', ',', '▁un', '▁ancien', '▁meurt', '▁et', '▁', 'emporte', '▁les', '▁dernières', '▁s', 'y', 'lla', 'be', 's', '▁avec', '▁lui', '▁dans', '▁la', '▁tombe', '▁d', "'", 'une', '▁langue', '▁ancienne', '.'] 2024-01-16 13:18:14,295 - DEBUG - joeynmt.training - Tokenized hypothesis: ['', '▁Comment', '▁ne', '▁pas', '▁se', '▁sentir', '▁seuls', ',', '▁en', 've', 'lop', 'pé', 's', '▁dans', '▁le', '▁silence', ',', '▁et', '▁être', '▁le', '▁dernier', '▁de', '▁votre', '▁peuple', '▁à', '▁parler', '▁votre', '▁langue', ',', '▁de', '▁n', "'", 'avoir', '▁aucun', '▁moyen', '▁de', '▁transmettre', '▁la', '▁sagesse', '▁des', '▁ancêtres', '▁ou', '▁de', 'van', 'c', 'er', '▁la', '▁promesse', '▁des', '▁enfants', '▁?', '', '▁Et', '▁pourtant', ',', '▁ce', '▁destin', '▁horrible', '▁est', '▁en', '▁fait', '▁le', '▁plaisir', '▁de', '▁quelqu', "'", 'un', '▁quelque', '▁part', '▁sur', '▁Terre', ',', '▁en', '▁gros', '▁tous', '▁les', '▁deux', '▁semaines', ',', '▁parce', '▁que', '▁tous', '▁les', '▁deux', '▁semaines', ',', '▁certains', '▁meurt', '▁et', '▁se', '▁porte', '▁avec', '▁lui', '▁dans', '▁la', '▁tombe', '▁le', '▁dernier', '▁s', 'y', 'll', 'able', '▁d', "'", 'une', '▁langue', '▁ancienne', '.'] 2024-01-16 13:18:14,295 - INFO - joeynmt.training - Example #3 2024-01-16 13:18:14,295 - INFO - joeynmt.training - Source: Du Kogi." 2024-01-16 13:18:14,295 - INFO - joeynmt.training - Reference: Let's make it Kogi." 2024-01-16 13:18:14,295 - INFO - joeynmt.training - Hypothesis: Du Kogi." 2024-01-16 13:18:14,296 - DEBUG - joeynmt.training - Tokenized source: ['▁Du', '▁Ko', 'g', 'i', '."'] 2024-01-16 13:18:14,296 - DEBUG - joeynmt.training - Tokenized reference: ['▁Let', "'", 's', '▁make', '▁it', '▁Ko', 'g', 'i', '."'] 2024-01-16 13:18:14,296 - DEBUG - joeynmt.training - Tokenized hypothesis: ['', '▁And', '▁I', '▁know', '▁there', "'", 's', '▁some', '▁of', '▁you', '▁who', '▁say', ',', '▁"', 'Well', ',', '▁would', 'n', "'", 't', '▁it', '▁be', '▁better', ',', '▁would', 'n', "'", 't', '▁the', '▁world', '▁be', '▁a', '▁better', '▁place', '▁if', '▁we', '▁all', '▁just', '▁spoke', '▁one', '▁language', '?"', '▁And', '▁I', '▁say', ',', '▁"', 'Great', ',', '▁let', "'", 's', '▁make', '▁that', '▁language', '▁Yoruba', '.', '▁Let', "'", 's', '▁make', '▁it', '▁Can', 'ton', 'es', 'e', '.', '', '▁Du', '▁Ko', 'g', 'i', '."'] 2024-01-16 13:19:31,219 - INFO - joeynmt.training - Epoch 6, Step: 15100, Batch Loss: 1.317333, Batch Acc: 0.687776, Tokens per Sec: 16279, Lr: 0.000163 2024-01-16 13:20:48,842 - INFO - joeynmt.training - Epoch 6, Step: 15200, Batch Loss: 1.275516, Batch Acc: 0.687046, Tokens per Sec: 16153, Lr: 0.000162 2024-01-16 13:22:05,216 - INFO - joeynmt.training - Epoch 6, Step: 15300, Batch Loss: 1.319065, Batch Acc: 0.689361, Tokens per Sec: 16351, Lr: 0.000162 2024-01-16 13:23:24,533 - INFO - joeynmt.training - Epoch 6, Step: 15400, Batch Loss: 1.307492, Batch Acc: 0.689355, Tokens per Sec: 15760, Lr: 0.000161 2024-01-16 13:24:41,207 - INFO - joeynmt.training - Epoch 6, Step: 15500, Batch Loss: 1.298629, Batch Acc: 0.688782, Tokens per Sec: 16330, Lr: 0.000161 2024-01-16 13:25:57,600 - INFO - joeynmt.training - Epoch 6, Step: 15600, Batch Loss: 1.273322, Batch Acc: 0.688424, Tokens per Sec: 16405, Lr: 0.000160 2024-01-16 13:27:13,515 - INFO - joeynmt.training - Epoch 6, Step: 15700, Batch Loss: 1.334399, Batch Acc: 0.689708, Tokens per Sec: 16468, Lr: 0.000160 2024-01-16 13:28:30,916 - INFO - joeynmt.training - Epoch 6, Step: 15800, Batch Loss: 1.380634, Batch Acc: 0.690190, Tokens per Sec: 16036, Lr: 0.000159 2024-01-16 13:29:47,420 - INFO - joeynmt.training - Epoch 6, Step: 15900, Batch Loss: 1.308075, Batch Acc: 0.691452, Tokens per Sec: 16268, Lr: 0.000159 2024-01-16 13:31:04,280 - INFO - joeynmt.training - Epoch 6, Step: 16000, Batch Loss: 1.303415, Batch Acc: 0.691409, Tokens per Sec: 16292, Lr: 0.000158 2024-01-16 13:31:04,327 - INFO - joeynmt.datasets - Sample random subset from dev data: n=500, seed=16042 2024-01-16 13:31:04,327 - INFO - joeynmt.prediction - Predicting 500 example(s)... (Greedy decoding with min_output_length=1, max_output_length=512, return_prob='none', generate_unk=True, repetition_penalty=-1, no_repeat_ngram_size=-1) 2024-01-16 13:31:18,396 - INFO - joeynmt.prediction - Generation took 14.0674[sec]. 2024-01-16 13:31:18,501 - INFO - joeynmt.metrics - nrefs:1|case:lc|eff:no|tok:13a|smooth:exp|version:2.4.0 2024-01-16 13:31:18,501 - INFO - joeynmt.prediction - Evaluation result (greedy): bleu: 26.89, loss: 2.01, ppl: 7.47, acc: 0.59, 0.0707[sec] 2024-01-16 13:31:21,605 - INFO - joeynmt.training - Checkpoint saved in models/iwslt14_prompt/16000.ckpt. 2024-01-16 13:31:21,607 - INFO - joeynmt.training - Example #0 2024-01-16 13:31:21,608 - INFO - joeynmt.training - Source: Nous procédons à des rites d'initiations. 2024-01-16 13:31:21,608 - INFO - joeynmt.training - Reference: We go through initiation rites. 2024-01-16 13:31:21,608 - INFO - joeynmt.training - Hypothesis: We're doing the English rites. 2024-01-16 13:31:21,609 - DEBUG - joeynmt.training - Tokenized source: ['▁Nous', '▁procéd', 'ons', '▁à', '▁des', '▁r', 'ites', '▁d', "'", 'ini', 'ti', 'ations', '.'] 2024-01-16 13:31:21,609 - DEBUG - joeynmt.training - Tokenized reference: ['▁We', '▁go', '▁through', '▁initiat', 'ion', '▁r', 'ites', '.'] 2024-01-16 13:31:21,609 - DEBUG - joeynmt.training - Tokenized hypothesis: ['', '▁We', "'", 're', '▁all', '▁born', '.', '▁We', '▁all', '▁bring', '▁our', '▁children', '▁into', '▁the', '▁world', '.', '', '▁We', "'", 're', '▁doing', '▁the', '▁English', '▁r', 'ites', '.'] 2024-01-16 13:31:21,609 - INFO - joeynmt.training - Example #1 2024-01-16 13:31:21,609 - INFO - joeynmt.training - Source: And this is an idea, if you think about it, can only fill you with hope. 2024-01-16 13:31:21,609 - INFO - joeynmt.training - Reference: Et ceci est une idée, si on y réfléchit, qui ne peut que vous remplir d'espoir. 2024-01-16 13:31:21,610 - INFO - joeynmt.training - Hypothesis: Et c'est une idée, si vous y réfléchissez, ne peut que vous remplir d'espoir. 2024-01-16 13:31:21,610 - DEBUG - joeynmt.training - Tokenized source: ['▁And', '▁this', '▁is', '▁an', '▁idea', ',', '▁if', '▁you', '▁think', '▁about', '▁it', ',', '▁can', '▁only', '▁fill', '▁you', '▁with', '▁hope', '.'] 2024-01-16 13:31:21,610 - DEBUG - joeynmt.training - Tokenized reference: ['▁Et', '▁ceci', '▁est', '▁une', '▁idée', ',', '▁si', '▁on', '▁y', '▁réfléchi', 't', ',', '▁qui', '▁ne', '▁peut', '▁que', '▁vous', '▁remplir', '▁d', "'", 'espoir', '.'] 2024-01-16 13:31:21,610 - DEBUG - joeynmt.training - Tokenized hypothesis: ['', '▁Tous', '▁ces', '▁peuple', 's', '▁nous', '▁enseigne', 'nt', '▁qu', "'", 'il', '▁y', '▁a', '▁d', "'", 'autres', '▁façons', '▁d', "'", 'être', ',', '▁d', "'", 'autres', '▁façons', '▁de', '▁penser', ',', '▁d', "'", 'autres', '▁manières', '▁de', '▁s', "'", 'orient', 'er', '▁sur', '▁Terre', '.', '', '▁Et', '▁c', "'", 'est', '▁une', '▁idée', ',', '▁si', '▁vous', '▁y', '▁réfléchissez', ',', '▁ne', '▁peut', '▁que', '▁vous', '▁remplir', '▁d', "'", 'espoir', '.'] 2024-01-16 13:31:21,610 - INFO - joeynmt.training - Example #2 2024-01-16 13:31:21,611 - INFO - joeynmt.training - Source: Et vous pourriez considérer ce tissu culturel de la vie en tant qu'ethnosphère et vous pourriez définir l'ethnosphère comme étant la somme globale de toutes les pensées, les rêves, les mythes, les idées, les inspirations, les intuitions engendrées par l'imagination humaine depuis l'aube de la conscience. 2024-01-16 13:31:21,611 - INFO - joeynmt.training - Reference: And you might think of this cultural web of life as being an ethnosphere, and you might define the ethnosphere as being the sum total of all thoughts and dreams, myths, ideas, inspirations, intuitions brought into being by the human imagination since the dawn of consciousness. 2024-01-16 13:31:21,611 - INFO - joeynmt.training - Hypothesis: And you might think of this cultural fabric of life as an athenospher and you might think of the as the global sum of all thoughts, dreams, myths, ideas, inspirations, intuitions that are caused by the human imagination since the dawn of consciousness. 2024-01-16 13:31:21,612 - DEBUG - joeynmt.training - Tokenized source: ['▁Et', '▁vous', '▁pourriez', '▁considérer', '▁ce', '▁tissu', '▁culturel', '▁de', '▁la', '▁vie', '▁en', '▁tant', '▁qu', "'", 'e', 'th', 'no', 's', 'ph', 'ère', '▁et', '▁vous', '▁pourriez', '▁définir', '▁l', "'", 'e', 'th', 'no', 's', 'ph', 'ère', '▁comme', '▁étant', '▁la', '▁somme', '▁globale', '▁de', '▁toutes', '▁les', '▁pensées', ',', '▁les', '▁rêves', ',', '▁les', '▁mythe', 's', ',', '▁les', '▁idées', ',', '▁les', '▁inspiration', 's', ',', '▁les', '▁intuition', 's', '▁engendré', 'es', '▁par', '▁l', "'", 'imagination', '▁humaine', '▁depuis', '▁l', "'", 'aube', '▁de', '▁la', '▁conscience', '.'] 2024-01-16 13:31:21,612 - DEBUG - joeynmt.training - Tokenized reference: ['▁And', '▁you', '▁might', '▁think', '▁of', '▁this', '▁cultural', '▁web', '▁of', '▁life', '▁as', '▁being', '▁an', '▁et', 'h', 'n', 'osphere', ',', '▁and', '▁you', '▁might', '▁define', '▁the', '▁et', 'h', 'n', 'osphere', '▁as', '▁being', '▁the', '▁sum', '▁total', '▁of', '▁all', '▁thoughts', '▁and', '▁dreams', ',', '▁myth', 's', ',', '▁ideas', ',', '▁inspiration', 's', ',', '▁intuition', 's', '▁brought', '▁into', '▁being', '▁by', '▁the', '▁human', '▁imagination', '▁since', '▁the', '▁dawn', '▁of', '▁consciousness', '.'] 2024-01-16 13:31:21,612 - DEBUG - joeynmt.training - Tokenized hypothesis: ['', '▁Now', ',', '▁together', '▁the', '▁myriad', '▁cultures', '▁of', '▁the', '▁world', '▁make', '▁up', '▁a', '▁web', '▁of', '▁spiritual', '▁life', '▁and', '▁cultural', '▁life', '▁that', '▁en', 've', 'lop', 's', '▁the', '▁planet', ',', '▁and', '▁is', '▁as', '▁important', '▁to', '▁the', '▁well', '-', 'being', '▁of', '▁the', '▁planet', '▁as', '▁indeed', '▁is', '▁the', '▁biological', '▁web', '▁of', '▁life', '▁that', '▁you', '▁know', '▁as', '▁a', '▁biosphere', '.', '', '▁And', '▁you', '▁might', '▁think', '▁of', '▁this', '▁cultural', '▁fabric', '▁of', '▁life', '▁as', '▁an', '▁a', 'the', 'no', 's', 'ph', 'er', '▁and', '▁you', '▁might', '▁think', '▁of', '▁the', '▁as', '▁the', '▁global', '▁sum', '▁of', '▁all', '▁thoughts', ',', '▁dreams', ',', '▁myth', 's', ',', '▁ideas', ',', '▁inspiration', 's', ',', '▁intuition', 's', '▁that', '▁are', '▁cause', 'd', '▁by', '▁the', '▁human', '▁imagination', '▁since', '▁the', '▁dawn', '▁of', '▁consciousness', '.'] 2024-01-16 13:31:21,612 - INFO - joeynmt.training - Example #3 2024-01-16 13:31:21,612 - INFO - joeynmt.training - Source: And the great indicator of that, of course, is language loss. 2024-01-16 13:31:21,612 - INFO - joeynmt.training - Reference: Et l'indicateur le plus fiable est bien sûr l'extinction du langage. 2024-01-16 13:31:21,612 - INFO - joeynmt.training - Hypothesis: Et le grand indicateur de cela, bien sûr, est la perte de la langue. 2024-01-16 13:31:21,613 - DEBUG - joeynmt.training - Tokenized source: ['▁And', '▁the', '▁great', '▁indicator', '▁of', '▁that', ',', '▁of', '▁course', ',', '▁is', '▁language', '▁loss', '.'] 2024-01-16 13:31:21,613 - DEBUG - joeynmt.training - Tokenized reference: ['▁Et', '▁l', "'", 'ind', 'ic', 'ateur', '▁le', '▁plus', '▁fiable', '▁est', '▁bien', '▁sûr', '▁l', "'", 'extinction', '▁du', '▁langage', '.'] 2024-01-16 13:31:21,613 - DEBUG - joeynmt.training - Tokenized hypothesis: ['', '▁Aucun', '▁biologist', 'e', ',', '▁par', '▁exemple', ',', '▁n', "'", 'ose', 'rait', '▁suggérer', '▁que', '▁50%', '▁ou', '▁plus', '▁de', '▁toutes', '▁les', '▁espèces', '▁ont', '▁été', '▁ou', '▁sont', '▁à', '▁deux', '▁doigts', '▁de', '▁l', "'", 'extinction', '▁parce', '▁que', '▁ce', '▁n', "'", 'est', '▁tout', '▁simplement', '▁pas', '▁vrai', ',', '▁et', '▁pourtant', '▁--', '▁que', '▁le', '▁scénario', '▁le', '▁plus', '▁a', 'po', 'ca', 'ly', 'p', 'tique', '▁dans', '▁le', '▁royaume', '▁de', '▁la', '▁diversité', '▁biologique', '▁--', '▁se', '▁rapproche', '▁rarement', '▁de', '▁ce', '▁que', '▁nous', '▁considér', 'ons', '▁comme', '▁le', '▁scénario', '▁le', '▁plus', '▁optimiste', '▁au', '▁sein', '▁de', '▁la', '▁diversité', '▁culturelle', '.', '', '▁Et', '▁le', '▁grand', '▁indicate', 'ur', '▁de', '▁cela', ',', '▁bien', '▁sûr', ',', '▁est', '▁la', '▁perte', '▁de', '▁la', '▁langue', '.'] 2024-01-16 13:32:38,073 - INFO - joeynmt.training - Epoch 6, Step: 16100, Batch Loss: 1.250836, Batch Acc: 0.691594, Tokens per Sec: 16347, Lr: 0.000158 2024-01-16 13:33:55,261 - INFO - joeynmt.training - Epoch 6, Step: 16200, Batch Loss: 1.262796, Batch Acc: 0.693704, Tokens per Sec: 16182, Lr: 0.000157 2024-01-16 13:35:12,015 - INFO - joeynmt.training - Epoch 6, Step: 16300, Batch Loss: 1.275432, Batch Acc: 0.692105, Tokens per Sec: 16303, Lr: 0.000157 2024-01-16 13:36:29,270 - INFO - joeynmt.training - Epoch 6, Step: 16400, Batch Loss: 1.269836, Batch Acc: 0.691842, Tokens per Sec: 16228, Lr: 0.000156 2024-01-16 13:37:14,144 - INFO - joeynmt.training - Epoch 6, total training loss: 3609.65, num. of seqs: 702202, num. of tokens: 34266040, 2107.4312[sec] 2024-01-16 13:37:14,154 - INFO - joeynmt.training - EPOCH 7 2024-01-16 13:37:46,284 - INFO - joeynmt.training - Epoch 7, Step: 16500, Batch Loss: 1.194453, Batch Acc: 0.708337, Tokens per Sec: 16364, Lr: 0.000156 2024-01-16 13:39:02,735 - INFO - joeynmt.training - Epoch 7, Step: 16600, Batch Loss: 1.173324, Batch Acc: 0.707387, Tokens per Sec: 16283, Lr: 0.000155 2024-01-16 13:40:18,714 - INFO - joeynmt.training - Epoch 7, Step: 16700, Batch Loss: 1.146339, Batch Acc: 0.709138, Tokens per Sec: 16394, Lr: 0.000155 2024-01-16 13:41:34,927 - INFO - joeynmt.training - Epoch 7, Step: 16800, Batch Loss: 1.222780, Batch Acc: 0.708564, Tokens per Sec: 16362, Lr: 0.000154 2024-01-16 13:42:51,880 - INFO - joeynmt.training - Epoch 7, Step: 16900, Batch Loss: 1.226919, Batch Acc: 0.707699, Tokens per Sec: 16222, Lr: 0.000154 2024-01-16 13:44:08,153 - INFO - joeynmt.training - Epoch 7, Step: 17000, Batch Loss: 1.163247, Batch Acc: 0.707977, Tokens per Sec: 16373, Lr: 0.000153 2024-01-16 13:44:08,154 - INFO - joeynmt.datasets - Sample random subset from dev data: n=500, seed=17042 2024-01-16 13:44:08,154 - INFO - joeynmt.prediction - Predicting 500 example(s)... (Greedy decoding with min_output_length=1, max_output_length=512, return_prob='none', generate_unk=True, repetition_penalty=-1, no_repeat_ngram_size=-1) 2024-01-16 13:44:22,764 - INFO - joeynmt.prediction - Generation took 14.6019[sec]. 2024-01-16 13:44:22,897 - INFO - joeynmt.metrics - nrefs:1|case:lc|eff:no|tok:13a|smooth:exp|version:2.4.0 2024-01-16 13:44:22,897 - INFO - joeynmt.prediction - Evaluation result (greedy): bleu: 28.32, loss: 2.04, ppl: 7.68, acc: 0.59, 0.1161[sec] 2024-01-16 13:44:22,898 - INFO - joeynmt.training - Hooray! New best validation result [bleu]! 2024-01-16 13:44:25,837 - INFO - joeynmt.training - Checkpoint saved in models/iwslt14_prompt/17000.ckpt. 2024-01-16 13:44:26,286 - INFO - joeynmt.training - Example #0 2024-01-16 13:44:26,287 - INFO - joeynmt.training - Source: We have to deal with the inexorable separation of death, so it shouldn't surprise us that we all sing, we all dance, we all have art. 2024-01-16 13:44:26,287 - INFO - joeynmt.training - Reference: Nous devons faire face à la séparation inexorable de la mort, cela ne devrait donc pas nous surprendre que nous chantions, nous dansions, nous sommes tous des artistes. 2024-01-16 13:44:26,287 - INFO - joeynmt.training - Hypothesis: Nous devons traiter la séparation inexorable de la mort, donc il ne faudrait pas nous surprendre que nous chantions tous, nous dansons tous, nous avons tous de l'art. 2024-01-16 13:44:26,288 - DEBUG - joeynmt.training - Tokenized source: ['▁We', '▁have', '▁to', '▁deal', '▁with', '▁the', '▁inexorable', '▁separation', '▁of', '▁death', ',', '▁so', '▁it', '▁should', 'n', "'", 't', '▁surprise', '▁us', '▁that', '▁we', '▁all', '▁sing', ',', '▁we', '▁all', '▁dance', ',', '▁we', '▁all', '▁have', '▁art', '.'] 2024-01-16 13:44:26,288 - DEBUG - joeynmt.training - Tokenized reference: ['▁Nous', '▁devons', '▁faire', '▁face', '▁à', '▁la', '▁séparation', '▁inexorable', '▁de', '▁la', '▁mort', ',', '▁cela', '▁ne', '▁devrait', '▁donc', '▁pas', '▁nous', '▁surprend', 're', '▁que', '▁nous', '▁chant', 'ions', ',', '▁nous', '▁dans', 'ions', ',', '▁nous', '▁sommes', '▁tous', '▁des', '▁artistes', '.'] 2024-01-16 13:44:26,288 - DEBUG - joeynmt.training - Tokenized hypothesis: ['', '▁Nous', '▁procéd', 'ons', '▁à', '▁des', '▁r', 'ites', '▁d', "'", 'ini', 'ti', 'ations', '.', '', '▁Nous', '▁devons', '▁traiter', '▁la', '▁séparation', '▁inexorable', '▁de', '▁la', '▁mort', ',', '▁donc', '▁il', '▁ne', '▁faudrait', '▁pas', '▁nous', '▁surprend', 're', '▁que', '▁nous', '▁chant', 'ions', '▁tous', ',', '▁nous', '▁dans', 'ons', '▁tous', ',', '▁nous', '▁avons', '▁tous', '▁de', '▁l', "'", 'art', '.'] 2024-01-16 13:44:26,288 - INFO - joeynmt.training - Example #1 2024-01-16 13:44:26,288 - INFO - joeynmt.training - Source: Et lorsque la biosphère fut sérieusement érodée, l'ethnosphère l'a été également -- et peut-être bien plus rapidement. 2024-01-16 13:44:26,288 - INFO - joeynmt.training - Reference: And just as the biosphere has been severely eroded, so too is the ethnosphere -- and, if anything, at a far greater rate. 2024-01-16 13:44:26,289 - INFO - joeynmt.training - Hypothesis: And when the biosphere was seriously degraded, the anethnospher was also -- and perhaps much more quickly. 2024-01-16 13:44:26,289 - DEBUG - joeynmt.training - Tokenized source: ['▁Et', '▁lorsque', '▁la', '▁biosphère', '▁fut', '▁sérieusement', '▁é', 'ro', 'd', 'ée', ',', '▁l', "'", 'e', 'th', 'no', 's', 'ph', 'ère', '▁l', "'", 'a', '▁été', '▁également', '▁--', '▁et', '▁peut', '-', 'être', '▁bien', '▁plus', '▁rapidement', '.'] 2024-01-16 13:44:26,289 - DEBUG - joeynmt.training - Tokenized reference: ['▁And', '▁just', '▁as', '▁the', '▁biosphere', '▁has', '▁been', '▁severely', '▁er', 'od', 'ed', ',', '▁so', '▁too', '▁is', '▁the', '▁et', 'h', 'n', 'osphere', '▁--', '▁and', ',', '▁if', '▁anything', ',', '▁at', '▁a', '▁far', '▁greater', '▁rate', '.'] 2024-01-16 13:44:26,289 - DEBUG - joeynmt.training - Tokenized hypothesis: ['', '▁It', "'", 's', '▁the', '▁symbol', '▁of', '▁all', '▁that', '▁we', '▁are', '▁and', '▁all', '▁that', '▁we', '▁can', '▁be', '▁as', '▁an', '▁astonishing', 'ly', '▁in', 'qui', 's', 'it', 'ive', '▁species', '.', '', '▁And', '▁when', '▁the', '▁biosphere', '▁was', '▁seriously', '▁degraded', ',', '▁the', '▁an', 'e', 'th', 'no', 's', 'ph', 'er', '▁was', '▁also', '▁--', '▁and', '▁perhaps', '▁much', '▁more', '▁quickly', '.'] 2024-01-16 13:44:26,290 - INFO - joeynmt.training - Example #2 2024-01-16 13:44:26,290 - INFO - joeynmt.training - Source: Chaque langue est une ancienne forêt de l'esprit, un partage, une pensée, un écosystème de possibilités spirituelles. 2024-01-16 13:44:26,290 - INFO - joeynmt.training - Reference: Every language is an old-growth forest of the mind, a watershed, a thought, an ecosystem of spiritual possibilities. 2024-01-16 13:44:26,290 - INFO - joeynmt.training - Hypothesis: Each language is an ancient forest of mind, a sharing, a thought, an ecosystem of spiritual possibilities. 2024-01-16 13:44:26,291 - DEBUG - joeynmt.training - Tokenized source: ['▁C', 'haque', '▁langue', '▁est', '▁une', '▁ancienne', '▁forêt', '▁de', '▁l', "'", 'esprit', ',', '▁un', '▁partage', ',', '▁une', '▁pensée', ',', '▁un', '▁écosystème', '▁de', '▁possibilités', '▁spirituelle', 's', '.'] 2024-01-16 13:44:26,291 - DEBUG - joeynmt.training - Tokenized reference: ['▁Every', '▁language', '▁is', '▁an', '▁old', '-', 'growth', '▁forest', '▁of', '▁the', '▁mind', ',', '▁a', '▁water', 'sh', 'ed', ',', '▁a', '▁thought', ',', '▁an', '▁ecosystem', '▁of', '▁spiritual', '▁possibilities', '.'] 2024-01-16 13:44:26,291 - DEBUG - joeynmt.training - Tokenized hypothesis: ['', '▁It', "'", 's', '▁a', '▁vehicle', '▁through', '▁which', '▁the', '▁soul', '▁of', '▁each', '▁particular', '▁culture', '▁comes', '▁into', '▁the', '▁material', '▁world', '.', '', '▁E', 'ach', '▁language', '▁is', '▁an', '▁ancient', '▁forest', '▁of', '▁mind', ',', '▁a', '▁sharing', ',', '▁a', '▁thought', ',', '▁an', '▁ecosystem', '▁of', '▁spiritual', '▁possibilities', '.'] 2024-01-16 13:44:26,291 - INFO - joeynmt.training - Example #3 2024-01-16 13:44:26,291 - INFO - joeynmt.training - Source: Et parmi ces 6,000 langues, alors que nous sommes à Monterey aujourd'hui, une bonne moitié n'est plus chuchotée dans les oreilles des enfants. 2024-01-16 13:44:26,291 - INFO - joeynmt.training - Reference: And of those 6,000 languages, as we sit here today in Monterey, fully half are no longer being whispered into the ears of children. 2024-01-16 13:44:26,292 - INFO - joeynmt.training - Hypothesis: And amongst those 6,000 languages, as we're in Monterey today, a good half is not whispered in the ears of children. 2024-01-16 13:44:26,292 - DEBUG - joeynmt.training - Tokenized source: ['▁Et', '▁parmi', '▁ces', '▁6,000', '▁langues', ',', '▁alors', '▁que', '▁nous', '▁sommes', '▁à', '▁Monterey', '▁aujourd', "'", 'hui', ',', '▁une', '▁bonne', '▁moitié', '▁n', "'", 'est', '▁plus', '▁ch', 'uch', 'ot', 'ée', '▁dans', '▁les', '▁oreilles', '▁des', '▁enfants', '.'] 2024-01-16 13:44:26,292 - DEBUG - joeynmt.training - Tokenized reference: ['▁And', '▁of', '▁those', '▁6,000', '▁languages', ',', '▁as', '▁we', '▁sit', '▁here', '▁today', '▁in', '▁Monterey', ',', '▁fully', '▁half', '▁are', '▁no', '▁longer', '▁being', '▁whisper', 'ed', '▁into', '▁the', '▁ears', '▁of', '▁children', '.'] 2024-01-16 13:44:26,293 - DEBUG - joeynmt.training - Tokenized hypothesis: ['', '▁Every', '▁language', '▁is', '▁an', '▁old', '-', 'growth', '▁forest', '▁of', '▁the', '▁mind', ',', '▁a', '▁water', 'sh', 'ed', ',', '▁a', '▁thought', ',', '▁an', '▁ecosystem', '▁of', '▁spiritual', '▁possibilities', '.', '', '▁And', '▁among', 'st', '▁those', '▁6,000', '▁languages', ',', '▁as', '▁we', "'", 're', '▁in', '▁Monterey', '▁today', ',', '▁a', '▁good', '▁half', '▁is', '▁not', '▁whisper', 'ed', '▁in', '▁the', '▁ears', '▁of', '▁children', '.'] 2024-01-16 13:45:41,858 - INFO - joeynmt.training - Epoch 7, Step: 17100, Batch Loss: 1.165600, Batch Acc: 0.708163, Tokens per Sec: 16493, Lr: 0.000153 2024-01-16 13:46:57,922 - INFO - joeynmt.training - Epoch 7, Step: 17200, Batch Loss: 1.166101, Batch Acc: 0.708850, Tokens per Sec: 16389, Lr: 0.000152 2024-01-16 13:48:13,265 - INFO - joeynmt.training - Epoch 7, Step: 17300, Batch Loss: 1.251985, Batch Acc: 0.708620, Tokens per Sec: 16614, Lr: 0.000152 2024-01-16 13:49:29,457 - INFO - joeynmt.training - Epoch 7, Step: 17400, Batch Loss: 1.269211, Batch Acc: 0.706234, Tokens per Sec: 16350, Lr: 0.000152 2024-01-16 13:50:44,328 - INFO - joeynmt.training - Epoch 7, Step: 17500, Batch Loss: 1.200514, Batch Acc: 0.707883, Tokens per Sec: 16638, Lr: 0.000151 2024-01-16 13:52:00,670 - INFO - joeynmt.training - Epoch 7, Step: 17600, Batch Loss: 1.135739, Batch Acc: 0.707026, Tokens per Sec: 16380, Lr: 0.000151 2024-01-16 13:53:16,900 - INFO - joeynmt.training - Epoch 7, Step: 17700, Batch Loss: 1.181727, Batch Acc: 0.708032, Tokens per Sec: 16395, Lr: 0.000150 2024-01-16 13:54:33,818 - INFO - joeynmt.training - Epoch 7, Step: 17800, Batch Loss: 1.185880, Batch Acc: 0.709671, Tokens per Sec: 16249, Lr: 0.000150 2024-01-16 13:55:50,792 - INFO - joeynmt.training - Epoch 7, Step: 17900, Batch Loss: 1.171691, Batch Acc: 0.709602, Tokens per Sec: 16324, Lr: 0.000149 2024-01-16 13:57:07,218 - INFO - joeynmt.training - Epoch 7, Step: 18000, Batch Loss: 1.176318, Batch Acc: 0.709508, Tokens per Sec: 16356, Lr: 0.000149 2024-01-16 13:57:07,219 - INFO - joeynmt.datasets - Sample random subset from dev data: n=500, seed=18042 2024-01-16 13:57:07,219 - INFO - joeynmt.prediction - Predicting 500 example(s)... (Greedy decoding with min_output_length=1, max_output_length=512, return_prob='none', generate_unk=True, repetition_penalty=-1, no_repeat_ngram_size=-1) 2024-01-16 13:57:35,233 - INFO - joeynmt.prediction - Generation took 28.0062[sec]. 2024-01-16 13:57:35,322 - INFO - joeynmt.metrics - nrefs:1|case:lc|eff:no|tok:13a|smooth:exp|version:2.4.0 2024-01-16 13:57:35,323 - INFO - joeynmt.prediction - Evaluation result (greedy): bleu: 26.55, loss: 2.04, ppl: 7.72, acc: 0.58, 0.0715[sec] 2024-01-16 13:57:38,282 - INFO - joeynmt.training - Checkpoint saved in models/iwslt14_prompt/18000.ckpt. 2024-01-16 13:57:38,691 - INFO - joeynmt.training - Example #0 2024-01-16 13:57:38,692 - INFO - joeynmt.training - Source: C'est le symbole de tout ce que nous sommes et tout ce que nous pouvons être en tant qu'espèce dotée d'une curiosité stupéfiante. 2024-01-16 13:57:38,692 - INFO - joeynmt.training - Reference: It's the symbol of all that we are and all that we can be as an astonishingly inquisitive species. 2024-01-16 13:57:38,692 - INFO - joeynmt.training - Hypothesis: It's the symbol of all of what we are and all of what we can be as a species that is really amazing curiosity. 2024-01-16 13:57:38,694 - DEBUG - joeynmt.training - Tokenized source: ['▁C', "'", 'est', '▁le', '▁symbole', '▁de', '▁tout', '▁ce', '▁que', '▁nous', '▁sommes', '▁et', '▁tout', '▁ce', '▁que', '▁nous', '▁pouvons', '▁être', '▁en', '▁tant', '▁qu', "'", 'espèce', '▁do', 't', 'ée', '▁d', "'", 'une', '▁curiosité', '▁stupéfiant', 'e', '.'] 2024-01-16 13:57:38,694 - DEBUG - joeynmt.training - Tokenized reference: ['▁It', "'", 's', '▁the', '▁symbol', '▁of', '▁all', '▁that', '▁we', '▁are', '▁and', '▁all', '▁that', '▁we', '▁can', '▁be', '▁as', '▁an', '▁astonishing', 'ly', '▁in', 'qui', 's', 'it', 'ive', '▁species', '.'] 2024-01-16 13:57:38,694 - DEBUG - joeynmt.training - Tokenized hypothesis: ['', '▁The', '▁et', 'h', 'n', 'osphere', '▁is', '▁humanity', "'", 's', '▁great', '▁legacy', '.', '', '▁It', "'", 's', '▁the', '▁symbol', '▁of', '▁all', '▁of', '▁what', '▁we', '▁are', '▁and', '▁all', '▁of', '▁what', '▁we', '▁can', '▁be', '▁as', '▁a', '▁species', '▁that', '▁is', '▁really', '▁amazing', '▁curiosity', '.'] 2024-01-16 13:57:38,694 - INFO - joeynmt.training - Example #1 2024-01-16 13:57:38,694 - INFO - joeynmt.training - Source: Et lorsque la biosphère fut sérieusement érodée, l'ethnosphère l'a été également -- et peut-être bien plus rapidement. 2024-01-16 13:57:38,694 - INFO - joeynmt.training - Reference: And just as the biosphere has been severely eroded, so too is the ethnosphere -- and, if anything, at a far greater rate. 2024-01-16 13:57:38,695 - INFO - joeynmt.training - Hypothesis: And when the biosphere was seriously erosed, the anthnospher was also -- and maybe much more quickly. 2024-01-16 13:57:38,696 - DEBUG - joeynmt.training - Tokenized source: ['▁Et', '▁lorsque', '▁la', '▁biosphère', '▁fut', '▁sérieusement', '▁é', 'ro', 'd', 'ée', ',', '▁l', "'", 'e', 'th', 'no', 's', 'ph', 'ère', '▁l', "'", 'a', '▁été', '▁également', '▁--', '▁et', '▁peut', '-', 'être', '▁bien', '▁plus', '▁rapidement', '.'] 2024-01-16 13:57:38,696 - DEBUG - joeynmt.training - Tokenized reference: ['▁And', '▁just', '▁as', '▁the', '▁biosphere', '▁has', '▁been', '▁severely', '▁er', 'od', 'ed', ',', '▁so', '▁too', '▁is', '▁the', '▁et', 'h', 'n', 'osphere', '▁--', '▁and', ',', '▁if', '▁anything', ',', '▁at', '▁a', '▁far', '▁greater', '▁rate', '.'] 2024-01-16 13:57:38,696 - DEBUG - joeynmt.training - Tokenized hypothesis: ['', '▁It', "'", 's', '▁the', '▁symbol', '▁of', '▁all', '▁that', '▁we', '▁are', '▁and', '▁all', '▁that', '▁we', '▁can', '▁be', '▁as', '▁an', '▁astonishing', 'ly', '▁in', 'qui', 's', 'it', 'ive', '▁species', '.', '', '▁And', '▁when', '▁the', '▁biosphere', '▁was', '▁seriously', '▁e', 'ro', 's', 'ed', ',', '▁the', '▁an', 'th', 'no', 's', 'ph', 'er', '▁was', '▁also', '▁--', '▁and', '▁maybe', '▁much', '▁more', '▁quickly', '.'] 2024-01-16 13:57:38,696 - INFO - joeynmt.training - Example #2 2024-01-16 13:57:38,696 - INFO - joeynmt.training - Source: No biologists, for example, would dare suggest that 50 percent of all species or more have been or are on the brink of extinction because it simply is not true, and yet that -- the most apocalyptic scenario in the realm of biological diversity -- scarcely approaches what we know to be the most optimistic scenario in the realm of cultural diversity. 2024-01-16 13:57:38,696 - INFO - joeynmt.training - Reference: Aucun biologiste, par exemple, n'oserait suggérer que 50% ou plus de toutes les espèces ont été ou sont à deux doigts de l'extinction parce que ce n'est tout simplement pas vrai, et pourtant -- que le scénario le plus apocalyptique dans le royaume de la diversité biologique -- se rapproche rarement de ce que nous considérons comme le scénario le plus optimiste au sein de la diversité culturelle. 2024-01-16 13:57:38,696 - INFO - joeynmt.training - Hypothesis: Pas de biologistes, par exemple, seraient en train de suggérer que 50 % de toutes les espèces ou plus ont été ou sont sur le bord de l'extinction parce que ce n'est tout simplement pas vrai, et pourtant cela -- le scénario le plus apocalyptiques dans le domaine de la diversité biologique -- approche rarement ce que nous savons être le scénario le plus optimiste dans le domaine de la diversité culturelle. 2024-01-16 13:57:38,698 - DEBUG - joeynmt.training - Tokenized source: ['▁No', '▁biologists', ',', '▁for', '▁example', ',', '▁would', '▁da', 're', '▁suggest', '▁that', '▁50', '▁percent', '▁of', '▁all', '▁species', '▁or', '▁more', '▁have', '▁been', '▁or', '▁are', '▁on', '▁the', '▁brin', 'k', '▁of', '▁extinction', '▁because', '▁it', '▁simply', '▁is', '▁not', '▁true', ',', '▁and', '▁yet', '▁that', '▁--', '▁the', '▁most', '▁apocalyptic', '▁scenario', '▁in', '▁the', '▁realm', '▁of', '▁biological', '▁diversity', '▁--', '▁scarce', 'ly', '▁approaches', '▁what', '▁we', '▁know', '▁to', '▁be', '▁the', '▁most', '▁optimistic', '▁scenario', '▁in', '▁the', '▁realm', '▁of', '▁cultural', '▁diversity', '.'] 2024-01-16 13:57:38,698 - DEBUG - joeynmt.training - Tokenized reference: ['▁Aucun', '▁biologist', 'e', ',', '▁par', '▁exemple', ',', '▁n', "'", 'ose', 'rait', '▁suggérer', '▁que', '▁50%', '▁ou', '▁plus', '▁de', '▁toutes', '▁les', '▁espèces', '▁ont', '▁été', '▁ou', '▁sont', '▁à', '▁deux', '▁doigts', '▁de', '▁l', "'", 'extinction', '▁parce', '▁que', '▁ce', '▁n', "'", 'est', '▁tout', '▁simplement', '▁pas', '▁vrai', ',', '▁et', '▁pourtant', '▁--', '▁que', '▁le', '▁scénario', '▁le', '▁plus', '▁a', 'po', 'ca', 'ly', 'p', 'tique', '▁dans', '▁le', '▁royaume', '▁de', '▁la', '▁diversité', '▁biologique', '▁--', '▁se', '▁rapproche', '▁rarement', '▁de', '▁ce', '▁que', '▁nous', '▁considér', 'ons', '▁comme', '▁le', '▁scénario', '▁le', '▁plus', '▁optimiste', '▁au', '▁sein', '▁de', '▁la', '▁diversité', '▁culturelle', '.'] 2024-01-16 13:57:38,698 - DEBUG - joeynmt.training - Tokenized hypothesis: ['', '▁Et', '▁lorsque', '▁la', '▁biosphère', '▁fut', '▁sérieusement', '▁é', 'ro', 'd', 'ée', ',', '▁l', "'", 'e', 'th', 'no', 's', 'ph', 'ère', '▁l', "'", 'a', '▁été', '▁également', '▁--', '▁et', '▁peut', '-', 'être', '▁bien', '▁plus', '▁rapidement', '.', '', '▁Pas', '▁de', '▁biologistes', ',', '▁par', '▁exemple', ',', '▁seraient', '▁en', '▁train', '▁de', '▁suggérer', '▁que', '▁50', '▁%', '▁de', '▁toutes', '▁les', '▁espèces', '▁ou', '▁plus', '▁ont', '▁été', '▁ou', '▁sont', '▁sur', '▁le', '▁bord', '▁de', '▁l', "'", 'extinction', '▁parce', '▁que', '▁ce', '▁n', "'", 'est', '▁tout', '▁simplement', '▁pas', '▁vrai', ',', '▁et', '▁pourtant', '▁cela', '▁--', '▁le', '▁scénario', '▁le', '▁plus', '▁a', 'po', 'ca', 'ly', 'p', 'tiques', '▁dans', '▁le', '▁domaine', '▁de', '▁la', '▁diversité', '▁biologique', '▁--', '▁approche', '▁rare', 'ment', '▁ce', '▁que', '▁nous', '▁savons', '▁être', '▁le', '▁scénario', '▁le', '▁plus', '▁optimiste', '▁dans', '▁le', '▁domaine', '▁de', '▁la', '▁diversité', '▁culturelle', '.'] 2024-01-16 13:57:38,698 - INFO - joeynmt.training - Example #3 2024-01-16 13:57:38,699 - INFO - joeynmt.training - Source: Chaque langue est une ancienne forêt de l'esprit, un partage, une pensée, un écosystème de possibilités spirituelles. 2024-01-16 13:57:38,699 - INFO - joeynmt.training - Reference: Every language is an old-growth forest of the mind, a watershed, a thought, an ecosystem of spiritual possibilities. 2024-01-16 13:57:38,699 - INFO - joeynmt.training - Hypothesis: Every language is an ancient forest of mind, a sharing, a thought, an ecosystem of spiritual possibilities. 2024-01-16 13:57:38,700 - DEBUG - joeynmt.training - Tokenized source: ['▁C', 'haque', '▁langue', '▁est', '▁une', '▁ancienne', '▁forêt', '▁de', '▁l', "'", 'esprit', ',', '▁un', '▁partage', ',', '▁une', '▁pensée', ',', '▁un', '▁écosystème', '▁de', '▁possibilités', '▁spirituelle', 's', '.'] 2024-01-16 13:57:38,700 - DEBUG - joeynmt.training - Tokenized reference: ['▁Every', '▁language', '▁is', '▁an', '▁old', '-', 'growth', '▁forest', '▁of', '▁the', '▁mind', ',', '▁a', '▁water', 'sh', 'ed', ',', '▁a', '▁thought', ',', '▁an', '▁ecosystem', '▁of', '▁spiritual', '▁possibilities', '.'] 2024-01-16 13:57:38,700 - DEBUG - joeynmt.training - Tokenized hypothesis: ['', '▁It', "'", 's', '▁a', '▁vehicle', '▁through', '▁which', '▁the', '▁soul', '▁of', '▁each', '▁particular', '▁culture', '▁comes', '▁into', '▁the', '▁material', '▁world', '.', '', '▁Every', '▁language', '▁is', '▁an', '▁ancient', '▁forest', '▁of', '▁mind', ',', '▁a', '▁sharing', ',', '▁a', '▁thought', ',', '▁an', '▁ecosystem', '▁of', '▁spiritual', '▁possibilities', '.'] 2024-01-16 13:58:55,726 - INFO - joeynmt.training - Epoch 7, Step: 18100, Batch Loss: 1.217816, Batch Acc: 0.708499, Tokens per Sec: 16220, Lr: 0.000149 2024-01-16 14:00:12,698 - INFO - joeynmt.training - Epoch 7, Step: 18200, Batch Loss: 1.195046, Batch Acc: 0.708862, Tokens per Sec: 16247, Lr: 0.000148 2024-01-16 14:01:29,400 - INFO - joeynmt.training - Epoch 7, Step: 18300, Batch Loss: 1.248441, Batch Acc: 0.708880, Tokens per Sec: 16290, Lr: 0.000148 2024-01-16 14:02:46,337 - INFO - joeynmt.training - Epoch 7, Step: 18400, Batch Loss: 1.254887, Batch Acc: 0.708206, Tokens per Sec: 16155, Lr: 0.000147 2024-01-16 14:04:02,918 - INFO - joeynmt.training - Epoch 7, Step: 18500, Batch Loss: 1.202551, Batch Acc: 0.708899, Tokens per Sec: 16307, Lr: 0.000147 2024-01-16 14:05:19,637 - INFO - joeynmt.training - Epoch 7, Step: 18600, Batch Loss: 1.157411, Batch Acc: 0.710251, Tokens per Sec: 16345, Lr: 0.000147 2024-01-16 14:06:36,436 - INFO - joeynmt.training - Epoch 7, Step: 18700, Batch Loss: 1.249651, Batch Acc: 0.710062, Tokens per Sec: 16189, Lr: 0.000146 2024-01-16 14:07:53,264 - INFO - joeynmt.training - Epoch 7, Step: 18800, Batch Loss: 1.240520, Batch Acc: 0.711011, Tokens per Sec: 16354, Lr: 0.000146 2024-01-16 14:09:10,963 - INFO - joeynmt.training - Epoch 7, Step: 18900, Batch Loss: 1.206951, Batch Acc: 0.709436, Tokens per Sec: 16143, Lr: 0.000145 2024-01-16 14:10:28,746 - INFO - joeynmt.training - Epoch 7, Step: 19000, Batch Loss: 1.222031, Batch Acc: 0.710157, Tokens per Sec: 16078, Lr: 0.000145 2024-01-16 14:10:28,778 - INFO - joeynmt.datasets - Sample random subset from dev data: n=500, seed=19042 2024-01-16 14:10:28,783 - INFO - joeynmt.prediction - Predicting 500 example(s)... (Greedy decoding with min_output_length=1, max_output_length=512, return_prob='none', generate_unk=True, repetition_penalty=-1, no_repeat_ngram_size=-1) 2024-01-16 14:10:47,794 - INFO - joeynmt.prediction - Generation took 18.8910[sec]. 2024-01-16 14:10:47,884 - INFO - joeynmt.metrics - nrefs:1|case:lc|eff:no|tok:13a|smooth:exp|version:2.4.0 2024-01-16 14:10:47,884 - INFO - joeynmt.prediction - Evaluation result (greedy): bleu: 27.41, loss: 2.02, ppl: 7.55, acc: 0.59, 0.0720[sec] 2024-01-16 14:10:50,945 - INFO - joeynmt.training - Checkpoint saved in models/iwslt14_prompt/19000.ckpt. 2024-01-16 14:10:51,358 - INFO - joeynmt.training - Example #0 2024-01-16 14:10:51,359 - INFO - joeynmt.training - Source: And of course, we all share the same adaptive imperatives. 2024-01-16 14:10:51,359 - INFO - joeynmt.training - Reference: Bien sûr, nous partageons tous les mêmes impératifs d'adaptation. 2024-01-16 14:10:51,359 - INFO - joeynmt.training - Hypothesis: Et bien sûr, nous partageons tous les mêmes impératifs adaptifs. 2024-01-16 14:10:51,360 - DEBUG - joeynmt.training - Tokenized source: ['▁And', '▁of', '▁course', ',', '▁we', '▁all', '▁share', '▁the', '▁same', '▁adaptive', '▁imperative', 's', '.'] 2024-01-16 14:10:51,360 - DEBUG - joeynmt.training - Tokenized reference: ['▁Bien', '▁sûr', ',', '▁nous', '▁partageons', '▁tous', '▁les', '▁mêmes', '▁im', 'pé', 'r', 'atif', 's', '▁d', "'", 'adaptation', '.'] 2024-01-16 14:10:51,360 - DEBUG - joeynmt.training - Tokenized hypothesis: ['', '▁Le', '▁fait', '▁de', '▁savoir', '▁que', '▁les', '▁Ja', 'gu', 'ar', '▁shaman', '▁voyage', 'nt', '▁toujours', '▁au', '-', 'delà', '▁de', '▁la', '▁voie', '▁l', 'act', 'ée', ',', '▁ou', '▁que', '▁les', '▁mythe', 's', '▁des', '▁anciens', '▁Inuit', '▁résonne', 'nt', '▁encore', '▁de', '▁sens', ',', '▁ou', '▁bien', '▁que', '▁dans', '▁l', "'", 'Himalaya', ',', '▁les', '▁Bo', 'ud', 'd', 'h', 'istes', '▁continuent', '▁à', '▁recherche', 'r', '▁le', '▁souffle', '▁du', '▁D', 'ha', 'r', 'ma', ',', '▁c', "'", 'est', '▁se', '▁rappeler', '▁de', '▁la', '▁révélation', '▁essentielle', '▁de', '▁l', "'", 'anthrop', 'ologie', ',', '▁et', '▁cela', '▁veut', '▁dire', '▁que', '▁le', '▁monde', '▁dans', '▁lequel', '▁nous', '▁vivons', '▁n', "'", 'existe', '▁pas', '▁dans', '▁un', '▁sens', '▁absolu', ',', '▁mais', '▁est', '▁uniquement', '▁un', '▁exemple', '▁de', '▁réalité', ',', '▁la', '▁conséquence', '▁d', "'", 'un', '▁ensemble', '▁spécifique', '▁de', '▁choix', '▁adapté', 's', '▁établi', 's', '▁par', '▁notre', '▁lignée', '▁avec', '▁succès', ',', '▁il', '▁y', '▁a', '▁plusieurs', '▁générations', '.', '', '▁Et', '▁bien', '▁sûr', ',', '▁nous', '▁partageons', '▁tous', '▁les', '▁mêmes', '▁im', 'pé', 'r', 'atif', 's', '▁adapt', 'if', 's', '.'] 2024-01-16 14:10:51,360 - INFO - joeynmt.training - Example #1 2024-01-16 14:10:51,361 - INFO - joeynmt.training - Source: All of these peoples teach us that there are other ways of being, other ways of thinking, other ways of orienting yourself in the Earth. 2024-01-16 14:10:51,361 - INFO - joeynmt.training - Reference: Tous ces peuples nous enseignent qu'il y a d'autres façons d'être, d'autres façons de penser, d'autres manières de s'orienter sur Terre. 2024-01-16 14:10:51,361 - INFO - joeynmt.training - Hypothesis: Toutes ces personnes nous apprennent qu'il y a d'autres façons de penser, d'autres façons de se orienter dans la Terre. 2024-01-16 14:10:51,361 - DEBUG - joeynmt.training - Tokenized source: ['▁All', '▁of', '▁these', '▁people', 's', '▁teach', '▁us', '▁that', '▁there', '▁are', '▁other', '▁ways', '▁of', '▁being', ',', '▁other', '▁ways', '▁of', '▁thinking', ',', '▁other', '▁ways', '▁of', '▁orient', 'ing', '▁yourself', '▁in', '▁the', '▁Earth', '.'] 2024-01-16 14:10:51,361 - DEBUG - joeynmt.training - Tokenized reference: ['▁Tous', '▁ces', '▁peuple', 's', '▁nous', '▁enseigne', 'nt', '▁qu', "'", 'il', '▁y', '▁a', '▁d', "'", 'autres', '▁façons', '▁d', "'", 'être', ',', '▁d', "'", 'autres', '▁façons', '▁de', '▁penser', ',', '▁d', "'", 'autres', '▁manières', '▁de', '▁s', "'", 'orient', 'er', '▁sur', '▁Terre', '.'] 2024-01-16 14:10:51,362 - DEBUG - joeynmt.training - Tokenized hypothesis: ['', '▁Et', '▁que', '▁ce', '▁soit', '▁le', '▁Pen', 'an', '▁dans', '▁les', '▁forêts', '▁du', '▁Bo', 'r', 'né', 'o', ',', '▁ou', '▁les', '▁a', 'co', 'ly', 'tes', '▁V', 'oodoo', '▁à', '▁Haïti', ',', '▁ou', '▁bien', '▁les', '▁guerrier', 's', '▁dans', '▁le', '▁désert', '▁du', '▁K', 'ais', 'ut', '▁au', '▁nord', '▁du', '▁Kenya', ',', '▁le', '▁C', 'ur', 'ende', 'ro', '▁dans', '▁les', '▁montagnes', '▁des', '▁And', 'es', ',', '▁ou', '▁bien', '▁un', '▁car', 'ava', 'n', 's', 'é', 'ra', 'il', '▁en', '▁plein', '▁milieu', '▁du', '▁Sahara', '.', '▁A', '▁propos', ',', '▁c', "'", 'est', '▁la', '▁personne', '▁avec', '▁qui', '▁j', "'", 'ai', '▁voyagé', '▁dans', '▁le', '▁désert', '▁il', '▁y', '▁un', '▁mois', ',', '▁ou', '▁effectivement', ',', '▁le', '▁gardien', '▁de', '▁troupeau', '▁de', '▁Yak', 's', '▁sur', '▁les', '▁fla', 'n', 'c', 's', '▁du', '▁Q', 'o', 'mo', 'lang', 'ma', ',', '▁l', "'", 'Everest', ',', '▁la', '▁dé', 'esse', '▁du', '▁monde', '.', '', '▁Toutes', '▁ces', '▁personnes', '▁nous', '▁apprennent', '▁qu', "'", 'il', '▁y', '▁a', '▁d', "'", 'autres', '▁façons', '▁de', '▁penser', ',', '▁d', "'", 'autres', '▁façons', '▁de', '▁se', '▁orient', 'er', '▁dans', '▁la', '▁Terre', '.'] 2024-01-16 14:10:51,362 - INFO - joeynmt.training - Example #2 2024-01-16 14:10:51,362 - INFO - joeynmt.training - Source: Lorsque chacun d'entre vous dans cette salle est né, 6,000 langues étaient pratiquées sur la planète. 2024-01-16 14:10:51,362 - INFO - joeynmt.training - Reference: When each of you in this room were born, there were 6,000 languages spoken on the planet. 2024-01-16 14:10:51,362 - INFO - joeynmt.training - Hypothesis: When each of you in this room were born, 6,000 languages were being practiced on the planet. 2024-01-16 14:10:51,363 - DEBUG - joeynmt.training - Tokenized source: ['▁Lorsque', '▁chacun', '▁d', "'", 'entre', '▁vous', '▁dans', '▁cette', '▁salle', '▁est', '▁né', ',', '▁6,000', '▁langues', '▁étaient', '▁pratiqu', 'ées', '▁sur', '▁la', '▁planète', '.'] 2024-01-16 14:10:51,363 - DEBUG - joeynmt.training - Tokenized reference: ['▁When', '▁each', '▁of', '▁you', '▁in', '▁this', '▁room', '▁were', '▁born', ',', '▁there', '▁were', '▁6,000', '▁languages', '▁spoken', '▁on', '▁the', '▁planet', '.'] 2024-01-16 14:10:51,363 - DEBUG - joeynmt.training - Tokenized hypothesis: ['', '▁And', '▁the', '▁great', '▁indicator', '▁of', '▁that', ',', '▁of', '▁course', ',', '▁is', '▁language', '▁loss', '.', '', '▁When', '▁each', '▁of', '▁you', '▁in', '▁this', '▁room', '▁were', '▁born', ',', '▁6,000', '▁languages', '▁were', '▁being', '▁practice', 'd', '▁on', '▁the', '▁planet', '.'] 2024-01-16 14:10:51,363 - INFO - joeynmt.training - Example #3 2024-01-16 14:10:51,363 - INFO - joeynmt.training - Source: Et parmi ces 6,000 langues, alors que nous sommes à Monterey aujourd'hui, une bonne moitié n'est plus chuchotée dans les oreilles des enfants. 2024-01-16 14:10:51,363 - INFO - joeynmt.training - Reference: And of those 6,000 languages, as we sit here today in Monterey, fully half are no longer being whispered into the ears of children. 2024-01-16 14:10:51,363 - INFO - joeynmt.training - Hypothesis: And amongst those 6.000 languages, as we're in Monterey today, a good half is no longer in the ears of the kids. 2024-01-16 14:10:51,364 - DEBUG - joeynmt.training - Tokenized source: ['▁Et', '▁parmi', '▁ces', '▁6,000', '▁langues', ',', '▁alors', '▁que', '▁nous', '▁sommes', '▁à', '▁Monterey', '▁aujourd', "'", 'hui', ',', '▁une', '▁bonne', '▁moitié', '▁n', "'", 'est', '▁plus', '▁ch', 'uch', 'ot', 'ée', '▁dans', '▁les', '▁oreilles', '▁des', '▁enfants', '.'] 2024-01-16 14:10:51,364 - DEBUG - joeynmt.training - Tokenized reference: ['▁And', '▁of', '▁those', '▁6,000', '▁languages', ',', '▁as', '▁we', '▁sit', '▁here', '▁today', '▁in', '▁Monterey', ',', '▁fully', '▁half', '▁are', '▁no', '▁longer', '▁being', '▁whisper', 'ed', '▁into', '▁the', '▁ears', '▁of', '▁children', '.'] 2024-01-16 14:10:51,364 - DEBUG - joeynmt.training - Tokenized hypothesis: ['', '▁Every', '▁language', '▁is', '▁an', '▁old', '-', 'growth', '▁forest', '▁of', '▁the', '▁mind', ',', '▁a', '▁water', 'sh', 'ed', ',', '▁a', '▁thought', ',', '▁an', '▁ecosystem', '▁of', '▁spiritual', '▁possibilities', '.', '', '▁And', '▁among', 'st', '▁those', '▁6.000', '▁languages', ',', '▁as', '▁we', "'", 're', '▁in', '▁Monterey', '▁today', ',', '▁a', '▁good', '▁half', '▁is', '▁no', '▁longer', '▁in', '▁the', '▁ears', '▁of', '▁the', '▁kids', '.'] 2024-01-16 14:12:09,134 - INFO - joeynmt.training - Epoch 7, Step: 19100, Batch Loss: 1.235157, Batch Acc: 0.710637, Tokens per Sec: 16136, Lr: 0.000145 2024-01-16 14:13:26,185 - INFO - joeynmt.training - Epoch 7, Step: 19200, Batch Loss: 1.191375, Batch Acc: 0.712838, Tokens per Sec: 16223, Lr: 0.000144 2024-01-16 14:13:26,927 - INFO - joeynmt.training - Epoch 7, total training loss: 3290.58, num. of seqs: 702202, num. of tokens: 34266040, 2100.2889[sec] 2024-01-16 14:13:26,937 - INFO - joeynmt.training - EPOCH 8 2024-01-16 14:14:42,407 - INFO - joeynmt.training - Epoch 8, Step: 19300, Batch Loss: 1.053337, Batch Acc: 0.728381, Tokens per Sec: 16408, Lr: 0.000144 2024-01-16 14:15:58,399 - INFO - joeynmt.training - Epoch 8, Step: 19400, Batch Loss: 1.071299, Batch Acc: 0.726327, Tokens per Sec: 16369, Lr: 0.000144 2024-01-16 14:17:14,849 - INFO - joeynmt.training - Epoch 8, Step: 19500, Batch Loss: 1.131594, Batch Acc: 0.728038, Tokens per Sec: 16350, Lr: 0.000143 2024-01-16 14:18:31,680 - INFO - joeynmt.training - Epoch 8, Step: 19600, Batch Loss: 1.117014, Batch Acc: 0.726884, Tokens per Sec: 16189, Lr: 0.000143 2024-01-16 14:19:48,081 - INFO - joeynmt.training - Epoch 8, Step: 19700, Batch Loss: 1.123798, Batch Acc: 0.725749, Tokens per Sec: 16300, Lr: 0.000142 2024-01-16 14:21:05,490 - INFO - joeynmt.training - Epoch 8, Step: 19800, Batch Loss: 1.091147, Batch Acc: 0.725395, Tokens per Sec: 16172, Lr: 0.000142 2024-01-16 14:22:22,343 - INFO - joeynmt.training - Epoch 8, Step: 19900, Batch Loss: 1.142316, Batch Acc: 0.725281, Tokens per Sec: 16252, Lr: 0.000142 2024-01-16 14:23:38,320 - INFO - joeynmt.training - Epoch 8, Step: 20000, Batch Loss: 1.063762, Batch Acc: 0.726826, Tokens per Sec: 16391, Lr: 0.000141 2024-01-16 14:23:38,321 - INFO - joeynmt.datasets - Sample random subset from dev data: n=500, seed=20042 2024-01-16 14:23:38,321 - INFO - joeynmt.prediction - Predicting 500 example(s)... (Greedy decoding with min_output_length=1, max_output_length=512, return_prob='none', generate_unk=True, repetition_penalty=-1, no_repeat_ngram_size=-1) 2024-01-16 14:23:53,392 - INFO - joeynmt.prediction - Generation took 15.0630[sec]. 2024-01-16 14:23:53,915 - INFO - joeynmt.metrics - nrefs:1|case:lc|eff:no|tok:13a|smooth:exp|version:2.4.0 2024-01-16 14:23:53,915 - INFO - joeynmt.prediction - Evaluation result (greedy): bleu: 27.44, loss: 1.97, ppl: 7.17, acc: 0.60, 0.1349[sec] 2024-01-16 14:23:56,941 - INFO - joeynmt.training - Checkpoint saved in models/iwslt14_prompt/20000.ckpt. 2024-01-16 14:23:57,384 - INFO - joeynmt.training - Example #0 2024-01-16 14:23:57,384 - INFO - joeynmt.training - Source: Et vous pourriez considérer ce tissu culturel de la vie en tant qu'ethnosphère et vous pourriez définir l'ethnosphère comme étant la somme globale de toutes les pensées, les rêves, les mythes, les idées, les inspirations, les intuitions engendrées par l'imagination humaine depuis l'aube de la conscience. 2024-01-16 14:23:57,384 - INFO - joeynmt.training - Reference: And you might think of this cultural web of life as being an ethnosphere, and you might define the ethnosphere as being the sum total of all thoughts and dreams, myths, ideas, inspirations, intuitions brought into being by the human imagination since the dawn of consciousness. 2024-01-16 14:23:57,385 - INFO - joeynmt.training - Hypothesis: And you might consider this cultural fabric of life as an aethnospher and you might consider the aethnospher as the overall sum of all thoughts, dreams, myths, ideas, inspirations, human imagination from the dawn of conscience. 2024-01-16 14:23:57,386 - DEBUG - joeynmt.training - Tokenized source: ['▁Et', '▁vous', '▁pourriez', '▁considérer', '▁ce', '▁tissu', '▁culturel', '▁de', '▁la', '▁vie', '▁en', '▁tant', '▁qu', "'", 'e', 'th', 'no', 's', 'ph', 'ère', '▁et', '▁vous', '▁pourriez', '▁définir', '▁l', "'", 'e', 'th', 'no', 's', 'ph', 'ère', '▁comme', '▁étant', '▁la', '▁somme', '▁globale', '▁de', '▁toutes', '▁les', '▁pensées', ',', '▁les', '▁rêves', ',', '▁les', '▁mythe', 's', ',', '▁les', '▁idées', ',', '▁les', '▁inspiration', 's', ',', '▁les', '▁intuition', 's', '▁engendré', 'es', '▁par', '▁l', "'", 'imagination', '▁humaine', '▁depuis', '▁l', "'", 'aube', '▁de', '▁la', '▁conscience', '.'] 2024-01-16 14:23:57,395 - DEBUG - joeynmt.training - Tokenized reference: ['▁And', '▁you', '▁might', '▁think', '▁of', '▁this', '▁cultural', '▁web', '▁of', '▁life', '▁as', '▁being', '▁an', '▁et', 'h', 'n', 'osphere', ',', '▁and', '▁you', '▁might', '▁define', '▁the', '▁et', 'h', 'n', 'osphere', '▁as', '▁being', '▁the', '▁sum', '▁total', '▁of', '▁all', '▁thoughts', '▁and', '▁dreams', ',', '▁myth', 's', ',', '▁ideas', ',', '▁inspiration', 's', ',', '▁intuition', 's', '▁brought', '▁into', '▁being', '▁by', '▁the', '▁human', '▁imagination', '▁since', '▁the', '▁dawn', '▁of', '▁consciousness', '.'] 2024-01-16 14:23:57,395 - DEBUG - joeynmt.training - Tokenized hypothesis: ['', '▁Now', ',', '▁together', '▁the', '▁myriad', '▁cultures', '▁of', '▁the', '▁world', '▁make', '▁up', '▁a', '▁web', '▁of', '▁spiritual', '▁life', '▁and', '▁cultural', '▁life', '▁that', '▁en', 've', 'lop', 's', '▁the', '▁planet', ',', '▁and', '▁is', '▁as', '▁important', '▁to', '▁the', '▁well', '-', 'being', '▁of', '▁the', '▁planet', '▁as', '▁indeed', '▁is', '▁the', '▁biological', '▁web', '▁of', '▁life', '▁that', '▁you', '▁know', '▁as', '▁a', '▁biosphere', '.', '', '▁And', '▁you', '▁might', '▁consider', '▁this', '▁cultural', '▁fabric', '▁of', '▁life', '▁as', '▁an', '▁a', 'e', 'th', 'no', 's', 'ph', 'er', '▁and', '▁you', '▁might', '▁consider', '▁the', '▁a', 'e', 'th', 'no', 's', 'ph', 'er', '▁as', '▁the', '▁overall', '▁sum', '▁of', '▁all', '▁thoughts', ',', '▁dreams', ',', '▁myth', 's', ',', '▁ideas', ',', '▁inspiration', 's', ',', '▁human', '▁imagination', '▁from', '▁the', '▁dawn', '▁of', '▁conscience', '.'] 2024-01-16 14:23:57,395 - INFO - joeynmt.training - Example #1 2024-01-16 14:23:57,396 - INFO - joeynmt.training - Source: It's the symbol of all that we are and all that we can be as an astonishingly inquisitive species. 2024-01-16 14:23:57,396 - INFO - joeynmt.training - Reference: C'est le symbole de tout ce que nous sommes et tout ce que nous pouvons être en tant qu'espèce dotée d'une curiosité stupéfiante. 2024-01-16 14:23:57,396 - INFO - joeynmt.training - Hypothesis: C'est le symbole de tout ce que nous sommes et de tout ce que nous pouvons être comme une espèce inquisitive étonnante. 2024-01-16 14:23:57,397 - DEBUG - joeynmt.training - Tokenized source: ['▁It', "'", 's', '▁the', '▁symbol', '▁of', '▁all', '▁that', '▁we', '▁are', '▁and', '▁all', '▁that', '▁we', '▁can', '▁be', '▁as', '▁an', '▁astonishing', 'ly', '▁in', 'qui', 's', 'it', 'ive', '▁species', '.'] 2024-01-16 14:23:57,397 - DEBUG - joeynmt.training - Tokenized reference: ['▁C', "'", 'est', '▁le', '▁symbole', '▁de', '▁tout', '▁ce', '▁que', '▁nous', '▁sommes', '▁et', '▁tout', '▁ce', '▁que', '▁nous', '▁pouvons', '▁être', '▁en', '▁tant', '▁qu', "'", 'espèce', '▁do', 't', 'ée', '▁d', "'", 'une', '▁curiosité', '▁stupéfiant', 'e', '.'] 2024-01-16 14:23:57,397 - DEBUG - joeynmt.training - Tokenized hypothesis: ['', '▁L', "'", 'e', 'th', 'no', 's', 'ph', 'ère', '▁est', '▁l', "'", 'héritage', '▁de', '▁l', "'", 'humanité', '.', '', '▁C', "'", 'est', '▁le', '▁symbole', '▁de', '▁tout', '▁ce', '▁que', '▁nous', '▁sommes', '▁et', '▁de', '▁tout', '▁ce', '▁que', '▁nous', '▁pouvons', '▁être', '▁comme', '▁une', '▁espèce', '▁in', 'qui', 's', 'it', 'ive', '▁étonnante', '.'] 2024-01-16 14:23:57,397 - INFO - joeynmt.training - Example #2 2024-01-16 14:23:57,397 - INFO - joeynmt.training - Source: Et l'indicateur le plus fiable est bien sûr l'extinction du langage. 2024-01-16 14:23:57,397 - INFO - joeynmt.training - Reference: And the great indicator of that, of course, is language loss. 2024-01-16 14:23:57,397 - INFO - joeynmt.training - Hypothesis: And the most reliable indicator is, of course, the extinction of language. 2024-01-16 14:23:57,398 - DEBUG - joeynmt.training - Tokenized source: ['▁Et', '▁l', "'", 'ind', 'ic', 'ateur', '▁le', '▁plus', '▁fiable', '▁est', '▁bien', '▁sûr', '▁l', "'", 'extinction', '▁du', '▁langage', '.'] 2024-01-16 14:23:57,398 - DEBUG - joeynmt.training - Tokenized reference: ['▁And', '▁the', '▁great', '▁indicator', '▁of', '▁that', ',', '▁of', '▁course', ',', '▁is', '▁language', '▁loss', '.'] 2024-01-16 14:23:57,398 - DEBUG - joeynmt.training - Tokenized hypothesis: ['', '▁No', '▁biologists', ',', '▁for', '▁example', ',', '▁would', '▁da', 're', '▁suggest', '▁that', '▁50', '▁percent', '▁of', '▁all', '▁species', '▁or', '▁more', '▁have', '▁been', '▁or', '▁are', '▁on', '▁the', '▁brin', 'k', '▁of', '▁extinction', '▁because', '▁it', '▁simply', '▁is', '▁not', '▁true', ',', '▁and', '▁yet', '▁that', '▁--', '▁the', '▁most', '▁apocalyptic', '▁scenario', '▁in', '▁the', '▁realm', '▁of', '▁biological', '▁diversity', '▁--', '▁scarce', 'ly', '▁approaches', '▁what', '▁we', '▁know', '▁to', '▁be', '▁the', '▁most', '▁optimistic', '▁scenario', '▁in', '▁the', '▁realm', '▁of', '▁cultural', '▁diversity', '.', '', '▁And', '▁the', '▁most', '▁reliable', '▁indicator', '▁is', ',', '▁of', '▁course', ',', '▁the', '▁extinction', '▁of', '▁language', '.'] 2024-01-16 14:23:57,398 - INFO - joeynmt.training - Example #3 2024-01-16 14:23:57,398 - INFO - joeynmt.training - Source: And of those 6,000 languages, as we sit here today in Monterey, fully half are no longer being whispered into the ears of children. 2024-01-16 14:23:57,399 - INFO - joeynmt.training - Reference: Et parmi ces 6,000 langues, alors que nous sommes à Monterey aujourd'hui, une bonne moitié n'est plus chuchotée dans les oreilles des enfants. 2024-01-16 14:23:57,399 - INFO - joeynmt.training - Hypothesis: Et de ces 6000 langues, comme nous sommes ici aujourd'hui à Monterey, la moitié ne sont plus chuchotées dans les oreilles d'enfants. 2024-01-16 14:23:57,399 - DEBUG - joeynmt.training - Tokenized source: ['▁And', '▁of', '▁those', '▁6,000', '▁languages', ',', '▁as', '▁we', '▁sit', '▁here', '▁today', '▁in', '▁Monterey', ',', '▁fully', '▁half', '▁are', '▁no', '▁longer', '▁being', '▁whisper', 'ed', '▁into', '▁the', '▁ears', '▁of', '▁children', '.'] 2024-01-16 14:23:57,399 - DEBUG - joeynmt.training - Tokenized reference: ['▁Et', '▁parmi', '▁ces', '▁6,000', '▁langues', ',', '▁alors', '▁que', '▁nous', '▁sommes', '▁à', '▁Monterey', '▁aujourd', "'", 'hui', ',', '▁une', '▁bonne', '▁moitié', '▁n', "'", 'est', '▁plus', '▁ch', 'uch', 'ot', 'ée', '▁dans', '▁les', '▁oreilles', '▁des', '▁enfants', '.'] 2024-01-16 14:23:57,399 - DEBUG - joeynmt.training - Tokenized hypothesis: ['', '▁C', 'haque', '▁langue', '▁est', '▁une', '▁ancienne', '▁forêt', '▁de', '▁l', "'", 'esprit', ',', '▁un', '▁partage', ',', '▁une', '▁pensée', ',', '▁un', '▁écosystème', '▁de', '▁possibilités', '▁spirituelle', 's', '.', '', '▁Et', '▁de', '▁ces', '▁6000', '▁langues', ',', '▁comme', '▁nous', '▁sommes', '▁ici', '▁aujourd', "'", 'hui', '▁à', '▁Monterey', ',', '▁la', '▁moitié', '▁ne', '▁sont', '▁plus', '▁ch', 'uch', 'ot', 'ées', '▁dans', '▁les', '▁oreilles', '▁d', "'", 'enfants', '.'] 2024-01-16 14:25:14,107 - INFO - joeynmt.training - Epoch 8, Step: 20100, Batch Loss: 1.036386, Batch Acc: 0.725548, Tokens per Sec: 16328, Lr: 0.000141 2024-01-16 14:26:30,890 - INFO - joeynmt.training - Epoch 8, Step: 20200, Batch Loss: 1.023703, Batch Acc: 0.724678, Tokens per Sec: 16257, Lr: 0.000141 2024-01-16 14:27:47,921 - INFO - joeynmt.training - Epoch 8, Step: 20300, Batch Loss: 1.150858, Batch Acc: 0.724838, Tokens per Sec: 16167, Lr: 0.000140 2024-01-16 14:29:05,347 - INFO - joeynmt.training - Epoch 8, Step: 20400, Batch Loss: 1.157100, Batch Acc: 0.724763, Tokens per Sec: 16167, Lr: 0.000140 2024-01-16 14:30:24,471 - INFO - joeynmt.training - Epoch 8, Step: 20500, Batch Loss: 1.118667, Batch Acc: 0.725516, Tokens per Sec: 15828, Lr: 0.000140 2024-01-16 14:31:40,766 - INFO - joeynmt.training - Epoch 8, Step: 20600, Batch Loss: 1.100182, Batch Acc: 0.725841, Tokens per Sec: 16430, Lr: 0.000139 2024-01-16 14:32:56,331 - INFO - joeynmt.training - Epoch 8, Step: 20700, Batch Loss: 1.113631, Batch Acc: 0.724787, Tokens per Sec: 16510, Lr: 0.000139 2024-01-16 14:34:11,632 - INFO - joeynmt.training - Epoch 8, Step: 20800, Batch Loss: 1.154153, Batch Acc: 0.726859, Tokens per Sec: 16506, Lr: 0.000139 2024-01-16 14:35:28,590 - INFO - joeynmt.training - Epoch 8, Step: 20900, Batch Loss: 1.112977, Batch Acc: 0.726893, Tokens per Sec: 16251, Lr: 0.000138 2024-01-16 14:36:45,532 - INFO - joeynmt.training - Epoch 8, Step: 21000, Batch Loss: 1.084469, Batch Acc: 0.725532, Tokens per Sec: 16330, Lr: 0.000138 2024-01-16 14:36:45,605 - INFO - joeynmt.datasets - Sample random subset from dev data: n=500, seed=21042 2024-01-16 14:36:45,622 - INFO - joeynmt.prediction - Predicting 500 example(s)... (Greedy decoding with min_output_length=1, max_output_length=512, return_prob='none', generate_unk=True, repetition_penalty=-1, no_repeat_ngram_size=-1) 2024-01-16 14:36:58,957 - INFO - joeynmt.prediction - Generation took 13.2944[sec]. 2024-01-16 14:36:59,084 - INFO - joeynmt.metrics - nrefs:1|case:lc|eff:no|tok:13a|smooth:exp|version:2.4.0 2024-01-16 14:36:59,085 - INFO - joeynmt.prediction - Evaluation result (greedy): bleu: 27.52, loss: 2.00, ppl: 7.40, acc: 0.59, 0.0720[sec] 2024-01-16 14:37:02,301 - INFO - joeynmt.training - Checkpoint saved in models/iwslt14_prompt/21000.ckpt. 2024-01-16 14:37:02,695 - INFO - joeynmt.training - Example #0 2024-01-16 14:37:02,696 - INFO - joeynmt.training - Source: The ethnosphere is humanity's great legacy. 2024-01-16 14:37:02,696 - INFO - joeynmt.training - Reference: L'ethnosphère est l'héritage de l'humanité. 2024-01-16 14:37:02,696 - INFO - joeynmt.training - Hypothesis: L'étosphäre est un grand héritage de l'humanité. 2024-01-16 14:37:02,697 - DEBUG - joeynmt.training - Tokenized source: ['▁The', '▁et', 'h', 'n', 'osphere', '▁is', '▁humanity', "'", 's', '▁great', '▁legacy', '.'] 2024-01-16 14:37:02,697 - DEBUG - joeynmt.training - Tokenized reference: ['▁L', "'", 'e', 'th', 'no', 's', 'ph', 'ère', '▁est', '▁l', "'", 'héritage', '▁de', '▁l', "'", 'humanité', '.'] 2024-01-16 14:37:02,697 - DEBUG - joeynmt.training - Tokenized hypothesis: ['', '▁Et', '▁vous', '▁pourriez', '▁considérer', '▁ce', '▁tissu', '▁culturel', '▁de', '▁la', '▁vie', '▁en', '▁tant', '▁qu', "'", 'e', 'th', 'no', 's', 'ph', 'ère', '▁et', '▁vous', '▁pourriez', '▁définir', '▁l', "'", 'e', 'th', 'no', 's', 'ph', 'ère', '▁comme', '▁étant', '▁la', '▁somme', '▁globale', '▁de', '▁toutes', '▁les', '▁pensées', ',', '▁les', '▁rêves', ',', '▁les', '▁mythe', 's', ',', '▁les', '▁idées', ',', '▁les', '▁inspiration', 's', ',', '▁les', '▁intuition', 's', '▁engendré', 'es', '▁par', '▁l', "'", 'imagination', '▁humaine', '▁depuis', '▁l', "'", 'aube', '▁de', '▁la', '▁conscience', '.', '', '▁L', "'", 'é', 't', 'o', 'sphäre', '▁est', '▁un', '▁grand', '▁héritage', '▁de', '▁l', "'", 'humanité', '.'] 2024-01-16 14:37:02,697 - INFO - joeynmt.training - Example #1 2024-01-16 14:37:02,698 - INFO - joeynmt.training - Source: And just as the biosphere has been severely eroded, so too is the ethnosphere -- and, if anything, at a far greater rate. 2024-01-16 14:37:02,698 - INFO - joeynmt.training - Reference: Et lorsque la biosphère fut sérieusement érodée, l'ethnosphère l'a été également -- et peut-être bien plus rapidement. 2024-01-16 14:37:02,698 - INFO - joeynmt.training - Hypothesis: Et tout comme la biosphère a été lourdement déchirée, la léosphäre aussi -- et, si quelque chose, à un taux bien plus important. 2024-01-16 14:37:02,699 - DEBUG - joeynmt.training - Tokenized source: ['▁And', '▁just', '▁as', '▁the', '▁biosphere', '▁has', '▁been', '▁severely', '▁er', 'od', 'ed', ',', '▁so', '▁too', '▁is', '▁the', '▁et', 'h', 'n', 'osphere', '▁--', '▁and', ',', '▁if', '▁anything', ',', '▁at', '▁a', '▁far', '▁greater', '▁rate', '.'] 2024-01-16 14:37:02,699 - DEBUG - joeynmt.training - Tokenized reference: ['▁Et', '▁lorsque', '▁la', '▁biosphère', '▁fut', '▁sérieusement', '▁é', 'ro', 'd', 'ée', ',', '▁l', "'", 'e', 'th', 'no', 's', 'ph', 'ère', '▁l', "'", 'a', '▁été', '▁également', '▁--', '▁et', '▁peut', '-', 'être', '▁bien', '▁plus', '▁rapidement', '.'] 2024-01-16 14:37:02,699 - DEBUG - joeynmt.training - Tokenized hypothesis: ['', '▁C', "'", 'est', '▁le', '▁symbole', '▁de', '▁tout', '▁ce', '▁que', '▁nous', '▁sommes', '▁et', '▁tout', '▁ce', '▁que', '▁nous', '▁pouvons', '▁être', '▁en', '▁tant', '▁qu', "'", 'espèce', '▁do', 't', 'ée', '▁d', "'", 'une', '▁curiosité', '▁stupéfiant', 'e', '.', '', '▁Et', '▁tout', '▁comme', '▁la', '▁biosphère', '▁a', '▁été', '▁lourde', 'ment', '▁dé', 'chi', 'r', 'ée', ',', '▁la', '▁l', 'é', 'o', 'sphäre', '▁aussi', '▁--', '▁et', ',', '▁si', '▁quelque', '▁chose', ',', '▁à', '▁un', '▁taux', '▁bien', '▁plus', '▁important', '.'] 2024-01-16 14:37:02,699 - INFO - joeynmt.training - Example #2 2024-01-16 14:37:02,699 - INFO - joeynmt.training - Source: Et lorsque la biosphère fut sérieusement érodée, l'ethnosphère l'a été également -- et peut-être bien plus rapidement. 2024-01-16 14:37:02,699 - INFO - joeynmt.training - Reference: And just as the biosphere has been severely eroded, so too is the ethnosphere -- and, if anything, at a far greater rate. 2024-01-16 14:37:02,699 - INFO - joeynmt.training - Hypothesis: And when the biosphere was seriously enclosed, the aethnospher was also -- and maybe much more quickly. 2024-01-16 14:37:02,700 - DEBUG - joeynmt.training - Tokenized source: ['▁Et', '▁lorsque', '▁la', '▁biosphère', '▁fut', '▁sérieusement', '▁é', 'ro', 'd', 'ée', ',', '▁l', "'", 'e', 'th', 'no', 's', 'ph', 'ère', '▁l', "'", 'a', '▁été', '▁également', '▁--', '▁et', '▁peut', '-', 'être', '▁bien', '▁plus', '▁rapidement', '.'] 2024-01-16 14:37:02,700 - DEBUG - joeynmt.training - Tokenized reference: ['▁And', '▁just', '▁as', '▁the', '▁biosphere', '▁has', '▁been', '▁severely', '▁er', 'od', 'ed', ',', '▁so', '▁too', '▁is', '▁the', '▁et', 'h', 'n', 'osphere', '▁--', '▁and', ',', '▁if', '▁anything', ',', '▁at', '▁a', '▁far', '▁greater', '▁rate', '.'] 2024-01-16 14:37:02,700 - DEBUG - joeynmt.training - Tokenized hypothesis: ['', '▁It', "'", 's', '▁the', '▁symbol', '▁of', '▁all', '▁that', '▁we', '▁are', '▁and', '▁all', '▁that', '▁we', '▁can', '▁be', '▁as', '▁an', '▁astonishing', 'ly', '▁in', 'qui', 's', 'it', 'ive', '▁species', '.', '', '▁And', '▁when', '▁the', '▁biosphere', '▁was', '▁seriously', '▁en', 'c', 'los', 'ed', ',', '▁the', '▁a', 'e', 'th', 'no', 's', 'ph', 'er', '▁was', '▁also', '▁--', '▁and', '▁maybe', '▁much', '▁more', '▁quickly', '.'] 2024-01-16 14:37:02,700 - INFO - joeynmt.training - Example #3 2024-01-16 14:37:02,701 - INFO - joeynmt.training - Source: C'est un véhicule à travers lequel l'âme de chaque culture spécifique entre dans le monde matériel. 2024-01-16 14:37:02,701 - INFO - joeynmt.training - Reference: It's a vehicle through which the soul of each particular culture comes into the material world. 2024-01-16 14:37:02,701 - INFO - joeynmt.training - Hypothesis: It's a vehicle through which the soul of every specific culture between the physical world. 2024-01-16 14:37:02,702 - DEBUG - joeynmt.training - Tokenized source: ['▁C', "'", 'est', '▁un', '▁véhicule', '▁à', '▁travers', '▁lequel', '▁l', "'", 'âme', '▁de', '▁chaque', '▁culture', '▁spécifique', '▁entre', '▁dans', '▁le', '▁monde', '▁matériel', '.'] 2024-01-16 14:37:02,702 - DEBUG - joeynmt.training - Tokenized reference: ['▁It', "'", 's', '▁a', '▁vehicle', '▁through', '▁which', '▁the', '▁soul', '▁of', '▁each', '▁particular', '▁culture', '▁comes', '▁into', '▁the', '▁material', '▁world', '.'] 2024-01-16 14:37:02,702 - DEBUG - joeynmt.training - Tokenized hypothesis: ['', '▁A', '▁language', '▁is', '▁a', '▁flash', '▁of', '▁the', '▁human', '▁spirit', '.', '', '▁It', "'", 's', '▁a', '▁vehicle', '▁through', '▁which', '▁the', '▁soul', '▁of', '▁every', '▁specific', '▁culture', '▁between', '▁the', '▁physical', '▁world', '.'] 2024-01-16 14:38:19,379 - INFO - joeynmt.training - Epoch 8, Step: 21100, Batch Loss: 1.140693, Batch Acc: 0.725353, Tokens per Sec: 16287, Lr: 0.000138 2024-01-16 14:39:36,197 - INFO - joeynmt.training - Epoch 8, Step: 21200, Batch Loss: 1.068338, Batch Acc: 0.725320, Tokens per Sec: 16329, Lr: 0.000137 2024-01-16 14:40:53,405 - INFO - joeynmt.training - Epoch 8, Step: 21300, Batch Loss: 1.131393, Batch Acc: 0.726668, Tokens per Sec: 16203, Lr: 0.000137 2024-01-16 14:42:10,177 - INFO - joeynmt.training - Epoch 8, Step: 21400, Batch Loss: 1.101503, Batch Acc: 0.726112, Tokens per Sec: 16332, Lr: 0.000137 2024-01-16 14:43:30,164 - INFO - joeynmt.training - Epoch 8, Step: 21500, Batch Loss: 1.080279, Batch Acc: 0.726761, Tokens per Sec: 15636, Lr: 0.000136 2024-01-16 14:44:47,940 - INFO - joeynmt.training - Epoch 8, Step: 21600, Batch Loss: 1.060910, Batch Acc: 0.726613, Tokens per Sec: 16112, Lr: 0.000136 2024-01-16 14:46:03,989 - INFO - joeynmt.training - Epoch 8, Step: 21700, Batch Loss: 1.083032, Batch Acc: 0.727814, Tokens per Sec: 16415, Lr: 0.000136 2024-01-16 14:47:20,441 - INFO - joeynmt.training - Epoch 8, Step: 21800, Batch Loss: 1.132197, Batch Acc: 0.727067, Tokens per Sec: 16345, Lr: 0.000135 2024-01-16 14:48:36,435 - INFO - joeynmt.training - Epoch 8, Step: 21900, Batch Loss: 1.105153, Batch Acc: 0.728022, Tokens per Sec: 16314, Lr: 0.000135 2024-01-16 14:49:10,009 - INFO - joeynmt.training - Epoch 8, total training loss: 3041.20, num. of seqs: 702202, num. of tokens: 34266040, 2106.7958[sec] 2024-01-16 14:49:10,072 - INFO - joeynmt.training - EPOCH 9 2024-01-16 14:49:52,739 - INFO - joeynmt.training - Epoch 9, Step: 22000, Batch Loss: 1.028307, Batch Acc: 0.745320, Tokens per Sec: 16420, Lr: 0.000135 2024-01-16 14:49:52,740 - INFO - joeynmt.datasets - Sample random subset from dev data: n=500, seed=22042 2024-01-16 14:49:52,740 - INFO - joeynmt.prediction - Predicting 500 example(s)... (Greedy decoding with min_output_length=1, max_output_length=512, return_prob='none', generate_unk=True, repetition_penalty=-1, no_repeat_ngram_size=-1) 2024-01-16 14:50:22,242 - INFO - joeynmt.prediction - Generation took 29.4939[sec]. 2024-01-16 14:50:22,380 - INFO - joeynmt.metrics - nrefs:1|case:lc|eff:no|tok:13a|smooth:exp|version:2.4.0 2024-01-16 14:50:22,380 - INFO - joeynmt.prediction - Evaluation result (greedy): bleu: 27.78, loss: 1.92, ppl: 6.82, acc: 0.60, 0.1203[sec] 2024-01-16 14:50:25,309 - INFO - joeynmt.training - Checkpoint saved in models/iwslt14_prompt/22000.ckpt. 2024-01-16 14:50:25,716 - INFO - joeynmt.training - Example #0 2024-01-16 14:50:25,718 - INFO - joeynmt.training - Source: Wade Davis sur les cultures en voie de disparition 2024-01-16 14:50:25,718 - INFO - joeynmt.training - Reference: Wade Davis: Dreams from endangered cultures 2024-01-16 14:50:25,718 - INFO - joeynmt.training - Hypothesis: Wade Davis's on the endangered crops. 2024-01-16 14:50:25,720 - DEBUG - joeynmt.training - Tokenized source: ['▁Wa', 'de', '▁Davis', '▁sur', '▁les', '▁cultures', '▁en', '▁voie', '▁de', '▁disparition'] 2024-01-16 14:50:25,720 - DEBUG - joeynmt.training - Tokenized reference: ['▁Wa', 'de', '▁Davis', ':', '▁Dream', 's', '▁from', '▁endangered', '▁cultures'] 2024-01-16 14:50:25,720 - DEBUG - joeynmt.training - Tokenized hypothesis: ['', '', '▁Wa', 'de', '▁Davis', "'", 's', '▁on', '▁the', '▁endangered', '▁crops', '.'] 2024-01-16 14:50:25,720 - INFO - joeynmt.training - Example #1 2024-01-16 14:50:25,720 - INFO - joeynmt.training - Source: Aujourd'hui, les innombrables cultures dans le monde constituent un tissu de vie spirituelle et culturelle qui enveloppe la planète, et qui est aussi important pour le bien-être de la planète que l'est également le tissu biologique de la vie que vous connaissez en tant que biosphère. 2024-01-16 14:50:25,721 - INFO - joeynmt.training - Reference: Now, together the myriad cultures of the world make up a web of spiritual life and cultural life that envelops the planet, and is as important to the well-being of the planet as indeed is the biological web of life that you know as a biosphere. 2024-01-16 14:50:25,721 - INFO - joeynmt.training - Hypothesis: Today, countless cultures around the world make up a fabric of spiritual and cultural life that wraps the planet, and that is as important for the well-being of the planet as it is also the biological fabric of life as you know it as a biosphere. 2024-01-16 14:50:25,722 - DEBUG - joeynmt.training - Tokenized source: ['▁A', 'ujourd', "'", 'hui', ',', '▁les', '▁', 'innombrables', '▁cultures', '▁dans', '▁le', '▁monde', '▁constituent', '▁un', '▁tissu', '▁de', '▁vie', '▁spirituelle', '▁et', '▁culturelle', '▁qui', '▁enveloppe', '▁la', '▁planète', ',', '▁et', '▁qui', '▁est', '▁aussi', '▁important', '▁pour', '▁le', '▁bien', '-', 'être', '▁de', '▁la', '▁planète', '▁que', '▁l', "'", 'est', '▁également', '▁le', '▁tissu', '▁biologique', '▁de', '▁la', '▁vie', '▁que', '▁vous', '▁connaissez', '▁en', '▁tant', '▁que', '▁biosphère', '.'] 2024-01-16 14:50:25,722 - DEBUG - joeynmt.training - Tokenized reference: ['▁Now', ',', '▁together', '▁the', '▁myriad', '▁cultures', '▁of', '▁the', '▁world', '▁make', '▁up', '▁a', '▁web', '▁of', '▁spiritual', '▁life', '▁and', '▁cultural', '▁life', '▁that', '▁en', 've', 'lop', 's', '▁the', '▁planet', ',', '▁and', '▁is', '▁as', '▁important', '▁to', '▁the', '▁well', '-', 'being', '▁of', '▁the', '▁planet', '▁as', '▁indeed', '▁is', '▁the', '▁biological', '▁web', '▁of', '▁life', '▁that', '▁you', '▁know', '▁as', '▁a', '▁biosphere', '.'] 2024-01-16 14:50:25,722 - DEBUG - joeynmt.training - Tokenized hypothesis: ['', '▁And', '▁this', '▁is', '▁an', '▁idea', ',', '▁if', '▁you', '▁think', '▁about', '▁it', ',', '▁can', '▁only', '▁fill', '▁you', '▁with', '▁hope', '.', '', '▁Today', ',', '▁countless', '▁cultures', '▁around', '▁the', '▁world', '▁make', '▁up', '▁a', '▁fabric', '▁of', '▁spiritual', '▁and', '▁cultural', '▁life', '▁that', '▁wrap', 's', '▁the', '▁planet', ',', '▁and', '▁that', '▁is', '▁as', '▁important', '▁for', '▁the', '▁well', '-', 'being', '▁of', '▁the', '▁planet', '▁as', '▁it', '▁is', '▁also', '▁the', '▁biological', '▁fabric', '▁of', '▁life', '▁as', '▁you', '▁know', '▁it', '▁as', '▁a', '▁biosphere', '.'] 2024-01-16 14:50:25,722 - INFO - joeynmt.training - Example #2 2024-01-16 14:50:25,723 - INFO - joeynmt.training - Source: And you might think of this cultural web of life as being an ethnosphere, and you might define the ethnosphere as being the sum total of all thoughts and dreams, myths, ideas, inspirations, intuitions brought into being by the human imagination since the dawn of consciousness. 2024-01-16 14:50:25,723 - INFO - joeynmt.training - Reference: Et vous pourriez considérer ce tissu culturel de la vie en tant qu'ethnosphère et vous pourriez définir l'ethnosphère comme étant la somme globale de toutes les pensées, les rêves, les mythes, les idées, les inspirations, les intuitions engendrées par l'imagination humaine depuis l'aube de la conscience. 2024-01-16 14:50:25,723 - INFO - joeynmt.training - Hypothesis: Et vous pourriez penser à ce web culturel de la vie comme étant une commode, et vous pourriez définir l'osphère ainsi que étant la somme totale de toutes les pensées et de rêves, des mythes, des idées, des intuitions, qui sont amenés par l'imagination humaine depuis l'aube de la conscience. 2024-01-16 14:50:25,724 - DEBUG - joeynmt.training - Tokenized source: ['▁And', '▁you', '▁might', '▁think', '▁of', '▁this', '▁cultural', '▁web', '▁of', '▁life', '▁as', '▁being', '▁an', '▁et', 'h', 'n', 'osphere', ',', '▁and', '▁you', '▁might', '▁define', '▁the', '▁et', 'h', 'n', 'osphere', '▁as', '▁being', '▁the', '▁sum', '▁total', '▁of', '▁all', '▁thoughts', '▁and', '▁dreams', ',', '▁myth', 's', ',', '▁ideas', ',', '▁inspiration', 's', ',', '▁intuition', 's', '▁brought', '▁into', '▁being', '▁by', '▁the', '▁human', '▁imagination', '▁since', '▁the', '▁dawn', '▁of', '▁consciousness', '.'] 2024-01-16 14:50:25,724 - DEBUG - joeynmt.training - Tokenized reference: ['▁Et', '▁vous', '▁pourriez', '▁considérer', '▁ce', '▁tissu', '▁culturel', '▁de', '▁la', '▁vie', '▁en', '▁tant', '▁qu', "'", 'e', 'th', 'no', 's', 'ph', 'ère', '▁et', '▁vous', '▁pourriez', '▁définir', '▁l', "'", 'e', 'th', 'no', 's', 'ph', 'ère', '▁comme', '▁étant', '▁la', '▁somme', '▁globale', '▁de', '▁toutes', '▁les', '▁pensées', ',', '▁les', '▁rêves', ',', '▁les', '▁mythe', 's', ',', '▁les', '▁idées', ',', '▁les', '▁inspiration', 's', ',', '▁les', '▁intuition', 's', '▁engendré', 'es', '▁par', '▁l', "'", 'imagination', '▁humaine', '▁depuis', '▁l', "'", 'aube', '▁de', '▁la', '▁conscience', '.'] 2024-01-16 14:50:25,724 - DEBUG - joeynmt.training - Tokenized hypothesis: ['', '▁A', 'ujourd', "'", 'hui', ',', '▁les', '▁', 'innombrables', '▁cultures', '▁dans', '▁le', '▁monde', '▁constituent', '▁un', '▁tissu', '▁de', '▁vie', '▁spirituelle', '▁et', '▁culturelle', '▁qui', '▁enveloppe', '▁la', '▁planète', ',', '▁et', '▁qui', '▁est', '▁aussi', '▁important', '▁pour', '▁le', '▁bien', '-', 'être', '▁de', '▁la', '▁planète', '▁que', '▁l', "'", 'est', '▁également', '▁le', '▁tissu', '▁biologique', '▁de', '▁la', '▁vie', '▁que', '▁vous', '▁connaissez', '▁en', '▁tant', '▁que', '▁biosphère', '.', '', '▁Et', '▁vous', '▁pourriez', '▁penser', '▁à', '▁ce', '▁web', '▁culturel', '▁de', '▁la', '▁vie', '▁comme', '▁étant', '▁une', '▁', 'com', 'm', 'ode', ',', '▁et', '▁vous', '▁pourriez', '▁définir', '▁l', "'", 'os', 'ph', 'ère', '▁ainsi', '▁que', '▁étant', '▁la', '▁somme', '▁totale', '▁de', '▁toutes', '▁les', '▁pensées', '▁et', '▁de', '▁rêves', ',', '▁des', '▁mythe', 's', ',', '▁des', '▁idées', ',', '▁des', '▁intuition', 's', ',', '▁qui', '▁sont', '▁amené', 's', '▁par', '▁l', "'", 'imagination', '▁humaine', '▁depuis', '▁l', "'", 'aube', '▁de', '▁la', '▁conscience', '.'] 2024-01-16 14:50:25,724 - INFO - joeynmt.training - Example #3 2024-01-16 14:50:25,725 - INFO - joeynmt.training - Source: When each of you in this room were born, there were 6,000 languages spoken on the planet. 2024-01-16 14:50:25,726 - INFO - joeynmt.training - Reference: Lorsque chacun d'entre vous dans cette salle est né, 6,000 langues étaient pratiquées sur la planète. 2024-01-16 14:50:25,726 - INFO - joeynmt.training - Hypothesis: Quand chacun d'entre vous ici est né, il y a eu 6000 langues parlées sur la planète. 2024-01-16 14:50:25,727 - DEBUG - joeynmt.training - Tokenized source: ['▁When', '▁each', '▁of', '▁you', '▁in', '▁this', '▁room', '▁were', '▁born', ',', '▁there', '▁were', '▁6,000', '▁languages', '▁spoken', '▁on', '▁the', '▁planet', '.'] 2024-01-16 14:50:25,727 - DEBUG - joeynmt.training - Tokenized reference: ['▁Lorsque', '▁chacun', '▁d', "'", 'entre', '▁vous', '▁dans', '▁cette', '▁salle', '▁est', '▁né', ',', '▁6,000', '▁langues', '▁étaient', '▁pratiqu', 'ées', '▁sur', '▁la', '▁planète', '.'] 2024-01-16 14:50:25,727 - DEBUG - joeynmt.training - Tokenized hypothesis: ['', '▁Et', '▁l', "'", 'ind', 'ic', 'ateur', '▁le', '▁plus', '▁fiable', '▁est', '▁bien', '▁sûr', '▁l', "'", 'extinction', '▁du', '▁langage', '.', '', '▁Quand', '▁chacun', '▁d', "'", 'entre', '▁vous', '▁ici', '▁est', '▁né', ',', '▁il', '▁y', '▁a', '▁eu', '▁6000', '▁langues', '▁parlé', 'es', '▁sur', '▁la', '▁planète', '.'] 2024-01-16 14:51:42,117 - INFO - joeynmt.training - Epoch 9, Step: 22100, Batch Loss: 1.013917, Batch Acc: 0.744273, Tokens per Sec: 16333, Lr: 0.000135 2024-01-16 14:52:58,432 - INFO - joeynmt.training - Epoch 9, Step: 22200, Batch Loss: 1.004947, Batch Acc: 0.742891, Tokens per Sec: 16393, Lr: 0.000134 2024-01-16 14:54:14,116 - INFO - joeynmt.training - Epoch 9, Step: 22300, Batch Loss: 1.023688, Batch Acc: 0.743314, Tokens per Sec: 16444, Lr: 0.000134 2024-01-16 14:55:30,105 - INFO - joeynmt.training - Epoch 9, Step: 22400, Batch Loss: 1.027240, Batch Acc: 0.742024, Tokens per Sec: 16369, Lr: 0.000134 2024-01-16 14:56:46,712 - INFO - joeynmt.training - Epoch 9, Step: 22500, Batch Loss: 1.007378, Batch Acc: 0.740555, Tokens per Sec: 16340, Lr: 0.000133 2024-01-16 14:58:02,899 - INFO - joeynmt.training - Epoch 9, Step: 22600, Batch Loss: 0.998270, Batch Acc: 0.740451, Tokens per Sec: 16362, Lr: 0.000133 2024-01-16 14:59:19,014 - INFO - joeynmt.training - Epoch 9, Step: 22700, Batch Loss: 1.042244, Batch Acc: 0.741414, Tokens per Sec: 16371, Lr: 0.000133 2024-01-16 15:00:35,979 - INFO - joeynmt.training - Epoch 9, Step: 22800, Batch Loss: 1.025231, Batch Acc: 0.740502, Tokens per Sec: 16304, Lr: 0.000132 2024-01-16 15:01:52,114 - INFO - joeynmt.training - Epoch 9, Step: 22900, Batch Loss: 1.029147, Batch Acc: 0.740543, Tokens per Sec: 16430, Lr: 0.000132 2024-01-16 15:03:09,358 - INFO - joeynmt.training - Epoch 9, Step: 23000, Batch Loss: 1.056174, Batch Acc: 0.741488, Tokens per Sec: 16223, Lr: 0.000132 2024-01-16 15:03:09,359 - INFO - joeynmt.datasets - Sample random subset from dev data: n=500, seed=23042 2024-01-16 15:03:09,360 - INFO - joeynmt.prediction - Predicting 500 example(s)... (Greedy decoding with min_output_length=1, max_output_length=512, return_prob='none', generate_unk=True, repetition_penalty=-1, no_repeat_ngram_size=-1) 2024-01-16 15:03:27,604 - INFO - joeynmt.prediction - Generation took 18.2367[sec]. 2024-01-16 15:03:27,690 - INFO - joeynmt.metrics - nrefs:1|case:lc|eff:no|tok:13a|smooth:exp|version:2.4.0 2024-01-16 15:03:27,690 - INFO - joeynmt.prediction - Evaluation result (greedy): bleu: 29.21, loss: 1.94, ppl: 6.93, acc: 0.60, 0.0695[sec] 2024-01-16 15:03:27,691 - INFO - joeynmt.training - Hooray! New best validation result [bleu]! 2024-01-16 15:03:30,713 - INFO - joeynmt.training - Checkpoint saved in models/iwslt14_prompt/23000.ckpt. 2024-01-16 15:03:30,717 - INFO - joeynmt.training - Example #0 2024-01-16 15:03:30,718 - INFO - joeynmt.training - Source: We're all born. We all bring our children into the world. 2024-01-16 15:03:30,718 - INFO - joeynmt.training - Reference: Nous sommes tous nés. Nous apportons nos enfants dans ce monde. 2024-01-16 15:03:30,718 - INFO - joeynmt.training - Hypothesis: Nous sommes tous nés. Nous avons tous mis nos enfants dans le monde. 2024-01-16 15:03:30,719 - DEBUG - joeynmt.training - Tokenized source: ['▁We', "'", 're', '▁all', '▁born', '.', '▁We', '▁all', '▁bring', '▁our', '▁children', '▁into', '▁the', '▁world', '.'] 2024-01-16 15:03:30,719 - DEBUG - joeynmt.training - Tokenized reference: ['▁Nous', '▁sommes', '▁tous', '▁né', 's', '.', '▁Nous', '▁apport', 'ons', '▁nos', '▁enfants', '▁dans', '▁ce', '▁monde', '.'] 2024-01-16 15:03:30,719 - DEBUG - joeynmt.training - Tokenized hypothesis: ['', '▁Bien', '▁sûr', ',', '▁nous', '▁partageons', '▁tous', '▁les', '▁mêmes', '▁im', 'pé', 'r', 'atif', 's', '▁d', "'", 'adaptation', '.', '', '▁Nous', '▁sommes', '▁tous', '▁né', 's', '.', '▁Nous', '▁avons', '▁tous', '▁mis', '▁nos', '▁enfants', '▁dans', '▁le', '▁monde', '.'] 2024-01-16 15:03:30,719 - INFO - joeynmt.training - Example #1 2024-01-16 15:03:30,719 - INFO - joeynmt.training - Source: And this is an idea, if you think about it, can only fill you with hope. 2024-01-16 15:03:30,720 - INFO - joeynmt.training - Reference: Et ceci est une idée, si on y réfléchit, qui ne peut que vous remplir d'espoir. 2024-01-16 15:03:30,720 - INFO - joeynmt.training - Hypothesis: Et c'est une idée, si vous y réfléchissez, qui ne peut vous remplir que d'espoir. 2024-01-16 15:03:30,720 - DEBUG - joeynmt.training - Tokenized source: ['▁And', '▁this', '▁is', '▁an', '▁idea', ',', '▁if', '▁you', '▁think', '▁about', '▁it', ',', '▁can', '▁only', '▁fill', '▁you', '▁with', '▁hope', '.'] 2024-01-16 15:03:30,720 - DEBUG - joeynmt.training - Tokenized reference: ['▁Et', '▁ceci', '▁est', '▁une', '▁idée', ',', '▁si', '▁on', '▁y', '▁réfléchi', 't', ',', '▁qui', '▁ne', '▁peut', '▁que', '▁vous', '▁remplir', '▁d', "'", 'espoir', '.'] 2024-01-16 15:03:30,720 - DEBUG - joeynmt.training - Tokenized hypothesis: ['', '▁Tous', '▁ces', '▁peuple', 's', '▁nous', '▁enseigne', 'nt', '▁qu', "'", 'il', '▁y', '▁a', '▁d', "'", 'autres', '▁façons', '▁d', "'", 'être', ',', '▁d', "'", 'autres', '▁façons', '▁de', '▁penser', ',', '▁d', "'", 'autres', '▁manières', '▁de', '▁s', "'", 'orient', 'er', '▁sur', '▁Terre', '.', '', '▁Et', '▁c', "'", 'est', '▁une', '▁idée', ',', '▁si', '▁vous', '▁y', '▁réfléchissez', ',', '▁qui', '▁ne', '▁peut', '▁vous', '▁remplir', '▁que', '▁d', "'", 'espoir', '.'] 2024-01-16 15:03:30,720 - INFO - joeynmt.training - Example #2 2024-01-16 15:03:30,721 - INFO - joeynmt.training - Source: Let's make it Kogi." 2024-01-16 15:03:30,721 - INFO - joeynmt.training - Reference: Du Kogi." 2024-01-16 15:03:30,721 - INFO - joeynmt.training - Hypothesis: Faisons-le Kogi." 2024-01-16 15:03:30,722 - DEBUG - joeynmt.training - Tokenized source: ['▁Let', "'", 's', '▁make', '▁it', '▁Ko', 'g', 'i', '."'] 2024-01-16 15:03:30,722 - DEBUG - joeynmt.training - Tokenized reference: ['▁Du', '▁Ko', 'g', 'i', '."'] 2024-01-16 15:03:30,722 - DEBUG - joeynmt.training - Tokenized hypothesis: ['', '▁Et', '▁je', '▁sais', '▁que', '▁certains', '▁d', "'", 'entre', '▁vous', '▁disent', ',', '▁"', 'Ne', '▁serait', '-', 'il', '▁pas', '▁mieux', '▁?', '▁Le', '▁monde', '▁ne', '▁serait', '-', 'il', '▁pas', '▁un', '▁meilleur', '▁endroit', '▁si', '▁nous', '▁ne', '▁parlions', '▁qu', "'", 'une', '▁seule', '▁langue', '▁?"', '▁Et', '▁je', '▁répond', 's', ',', '▁"', 'Bien', ',', '▁cette', '▁langue', '▁sera', '▁du', '▁Yoruba', '.', '▁Du', '▁Can', 'ton', 'ais', '.', '', '▁Faisons', '-', 'le', '▁Ko', 'g', 'i', '."'] 2024-01-16 15:03:30,722 - INFO - joeynmt.training - Example #3 2024-01-16 15:03:30,722 - INFO - joeynmt.training - Source: Beaucoup d'entre nous oublient un peu que lorsque je dis "des façons différentes d'être", je veux vraiment dire des façons différentes d'être. 2024-01-16 15:03:30,722 - INFO - joeynmt.training - Reference: Now, there are many of us who sort of forget that when I say "different ways of being," I really do mean different ways of being. 2024-01-16 15:03:30,722 - INFO - joeynmt.training - Hypothesis: Many of us forget a little bit when I say, "In different ways to be," I really mean different ways of being. 2024-01-16 15:03:30,723 - DEBUG - joeynmt.training - Tokenized source: ['▁Beaucoup', '▁d', "'", 'entre', '▁nous', '▁', 'oubli', 'ent', '▁un', '▁peu', '▁que', '▁lorsque', '▁je', '▁dis', '▁"', 'des', '▁façons', '▁différentes', '▁d', "'", 'être', '",', '▁je', '▁veux', '▁vraiment', '▁dire', '▁des', '▁façons', '▁différentes', '▁d', "'", 'être', '.'] 2024-01-16 15:03:30,723 - DEBUG - joeynmt.training - Tokenized reference: ['▁Now', ',', '▁there', '▁are', '▁many', '▁of', '▁us', '▁who', '▁sort', '▁of', '▁forget', '▁that', '▁when', '▁I', '▁say', '▁"', 'd', 'iff', 'er', 'ent', '▁ways', '▁of', '▁being', ',"', '▁I', '▁really', '▁do', '▁mean', '▁different', '▁ways', '▁of', '▁being', '.'] 2024-01-16 15:03:30,723 - DEBUG - joeynmt.training - Tokenized hypothesis: ['', '▁And', '▁so', ',', '▁what', '▁I', "'", 'd', '▁like', '▁to', '▁do', '▁with', '▁you', '▁today', '▁is', '▁sort', '▁of', '▁take', '▁you', '▁on', '▁a', '▁journey', '▁through', '▁the', '▁et', 'h', 'n', 'osphere', ',', '▁a', '▁brief', '▁journey', '▁through', '▁the', '▁et', 'h', 'n', 'osphere', ',', '▁to', '▁try', '▁to', '▁begin', '▁to', '▁give', '▁you', '▁a', '▁sense', '▁of', '▁what', '▁in', '▁fact', '▁is', '▁being', '▁lost', '.', '', '▁Many', '▁of', '▁us', '▁forget', '▁a', '▁little', '▁bit', '▁when', '▁I', '▁say', ',', '▁"', 'In', '▁different', '▁ways', '▁to', '▁be', ',"', '▁I', '▁really', '▁mean', '▁different', '▁ways', '▁of', '▁being', '.'] 2024-01-16 15:04:46,739 - INFO - joeynmt.training - Epoch 9, Step: 23100, Batch Loss: 1.028367, Batch Acc: 0.741321, Tokens per Sec: 16450, Lr: 0.000132 2024-01-16 15:06:03,274 - INFO - joeynmt.training - Epoch 9, Step: 23200, Batch Loss: 1.031541, Batch Acc: 0.740135, Tokens per Sec: 16327, Lr: 0.000131 2024-01-16 15:07:19,200 - INFO - joeynmt.training - Epoch 9, Step: 23300, Batch Loss: 1.056157, Batch Acc: 0.741107, Tokens per Sec: 16477, Lr: 0.000131 2024-01-16 15:08:35,933 - INFO - joeynmt.training - Epoch 9, Step: 23400, Batch Loss: 0.989116, Batch Acc: 0.740336, Tokens per Sec: 16226, Lr: 0.000131 2024-01-16 15:09:53,086 - INFO - joeynmt.training - Epoch 9, Step: 23500, Batch Loss: 1.058120, Batch Acc: 0.741583, Tokens per Sec: 16174, Lr: 0.000130 2024-01-16 15:11:11,183 - INFO - joeynmt.training - Epoch 9, Step: 23600, Batch Loss: 1.063228, Batch Acc: 0.740312, Tokens per Sec: 16036, Lr: 0.000130 2024-01-16 15:12:27,926 - INFO - joeynmt.training - Epoch 9, Step: 23700, Batch Loss: 1.013027, Batch Acc: 0.741351, Tokens per Sec: 16256, Lr: 0.000130 2024-01-16 15:13:44,468 - INFO - joeynmt.training - Epoch 9, Step: 23800, Batch Loss: 1.052978, Batch Acc: 0.740222, Tokens per Sec: 16385, Lr: 0.000130 2024-01-16 15:15:02,720 - INFO - joeynmt.training - Epoch 9, Step: 23900, Batch Loss: 1.080914, Batch Acc: 0.739646, Tokens per Sec: 15971, Lr: 0.000129 2024-01-16 15:16:19,811 - INFO - joeynmt.training - Epoch 9, Step: 24000, Batch Loss: 1.042285, Batch Acc: 0.739051, Tokens per Sec: 16213, Lr: 0.000129 2024-01-16 15:16:19,903 - INFO - joeynmt.datasets - Sample random subset from dev data: n=500, seed=24042 2024-01-16 15:16:19,908 - INFO - joeynmt.prediction - Predicting 500 example(s)... (Greedy decoding with min_output_length=1, max_output_length=512, return_prob='none', generate_unk=True, repetition_penalty=-1, no_repeat_ngram_size=-1) 2024-01-16 15:16:36,222 - INFO - joeynmt.prediction - Generation took 16.2686[sec]. 2024-01-16 15:16:36,308 - INFO - joeynmt.metrics - nrefs:1|case:lc|eff:no|tok:13a|smooth:exp|version:2.4.0 2024-01-16 15:16:36,308 - INFO - joeynmt.prediction - Evaluation result (greedy): bleu: 27.79, loss: 1.96, ppl: 7.07, acc: 0.60, 0.0672[sec] 2024-01-16 15:16:39,441 - INFO - joeynmt.training - Checkpoint saved in models/iwslt14_prompt/24000.ckpt. 2024-01-16 15:16:39,444 - INFO - joeynmt.training - Example #0 2024-01-16 15:16:40,261 - INFO - joeynmt.training - Source: Le fait de savoir que les Jaguar shaman voyagent toujours au-delà de la voie lactée, ou que les mythes des anciens Inuit résonnent encore de sens, ou bien que dans l'Himalaya, les Bouddhistes continuent à rechercher le souffle du Dharma, c'est se rappeler de la révélation essentielle de l'anthropologie, et cela veut dire que le monde dans lequel nous vivons n'existe pas dans un sens absolu, mais est uniquement un exemple de réalité, la conséquence d'un ensemble spécifique de choix adaptés établis par notre lignée avec succès, il y a plusieurs générations. 2024-01-16 15:16:40,261 - INFO - joeynmt.training - Reference: Just to know that Jaguar shamans still journey beyond the Milky Way, or the myths of the Inuit elders still resonate with meaning, or that in the Himalaya, the Buddhists still pursue the breath of the Dharma, is to really remember the central revelation of anthropology, and that is the idea that the world in which we live does not exist in some absolute sense, but is just one model of reality, the consequence of one particular set of adaptive choices that our lineage made, albeit successfully, many generations ago. 2024-01-16 15:16:40,262 - INFO - joeynmt.training - Hypothesis: The fact of knowing that the shaman's Jaguar still travels beyond the lacted path, or the myths of the ancient Inuit still resonate, or that in the Himalayas, the Buddhists continue to search for the breath of Dharma, is to remember the basic insight of anthropology, and that means that the world in which we live doesn't exist in an absolute sense, but is just an example of a consequence of a specific set of choices that our succession has several generations. 2024-01-16 15:16:40,263 - DEBUG - joeynmt.training - Tokenized source: ['▁Le', '▁fait', '▁de', '▁savoir', '▁que', '▁les', '▁Ja', 'gu', 'ar', '▁shaman', '▁voyage', 'nt', '▁toujours', '▁au', '-', 'delà', '▁de', '▁la', '▁voie', '▁l', 'act', 'ée', ',', '▁ou', '▁que', '▁les', '▁mythe', 's', '▁des', '▁anciens', '▁Inuit', '▁résonne', 'nt', '▁encore', '▁de', '▁sens', ',', '▁ou', '▁bien', '▁que', '▁dans', '▁l', "'", 'Himalaya', ',', '▁les', '▁Bo', 'ud', 'd', 'h', 'istes', '▁continuent', '▁à', '▁recherche', 'r', '▁le', '▁souffle', '▁du', '▁D', 'ha', 'r', 'ma', ',', '▁c', "'", 'est', '▁se', '▁rappeler', '▁de', '▁la', '▁révélation', '▁essentielle', '▁de', '▁l', "'", 'anthrop', 'ologie', ',', '▁et', '▁cela', '▁veut', '▁dire', '▁que', '▁le', '▁monde', '▁dans', '▁lequel', '▁nous', '▁vivons', '▁n', "'", 'existe', '▁pas', '▁dans', '▁un', '▁sens', '▁absolu', ',', '▁mais', '▁est', '▁uniquement', '▁un', '▁exemple', '▁de', '▁réalité', ',', '▁la', '▁conséquence', '▁d', "'", 'un', '▁ensemble', '▁spécifique', '▁de', '▁choix', '▁adapté', 's', '▁établi', 's', '▁par', '▁notre', '▁lignée', '▁avec', '▁succès', ',', '▁il', '▁y', '▁a', '▁plusieurs', '▁générations', '.'] 2024-01-16 15:16:40,263 - DEBUG - joeynmt.training - Tokenized reference: ['▁Just', '▁to', '▁know', '▁that', '▁Ja', 'gu', 'ar', '▁shaman', 's', '▁still', '▁journey', '▁beyond', '▁the', '▁Milk', 'y', '▁Way', ',', '▁or', '▁the', '▁myth', 's', '▁of', '▁the', '▁Inuit', '▁el', 'der', 's', '▁still', '▁resonate', '▁with', '▁meaning', ',', '▁or', '▁that', '▁in', '▁the', '▁Himalaya', ',', '▁the', '▁Buddhist', 's', '▁still', '▁pursue', '▁the', '▁breath', '▁of', '▁the', '▁D', 'ha', 'r', 'ma', ',', '▁is', '▁to', '▁really', '▁remember', '▁the', '▁central', '▁revelation', '▁of', '▁anthropology', ',', '▁and', '▁that', '▁is', '▁the', '▁idea', '▁that', '▁the', '▁world', '▁in', '▁which', '▁we', '▁live', '▁does', '▁not', '▁exist', '▁in', '▁some', '▁absolute', '▁sense', ',', '▁but', '▁is', '▁just', '▁one', '▁model', '▁of', '▁reality', ',', '▁the', '▁consequence', '▁of', '▁one', '▁particular', '▁set', '▁of', '▁adaptive', '▁choices', '▁that', '▁our', '▁line', 'age', '▁made', ',', '▁albeit', '▁successfully', ',', '▁many', '▁generations', '▁ago', '.'] 2024-01-16 15:16:40,263 - DEBUG - joeynmt.training - Tokenized hypothesis: ['', '▁You', '▁know', ',', '▁one', '▁of', '▁the', '▁intense', '▁pleasure', 's', '▁of', '▁travel', '▁and', '▁one', '▁of', '▁the', '▁de', 'light', 's', '▁of', '▁et', 'h', 'n', 'ographic', '▁research', '▁is', '▁the', '▁opportunity', '▁to', '▁live', '▁among', 'st', '▁those', '▁who', '▁have', '▁not', '▁forgotten', '▁the', '▁old', '▁ways', ',', '▁who', '▁still', '▁feel', '▁their', '▁past', '▁in', '▁the', '▁wind', ',', '▁touch', '▁it', '▁in', '▁stones', '▁polish', 'ed', '▁by', '▁rain', ',', '▁taste', '▁it', '▁in', '▁the', '▁bitter', '▁leaves', '▁of', '▁plants', '.', '', '▁The', '▁fact', '▁of', '▁knowing', '▁that', '▁the', '▁shaman', "'", 's', '▁Ja', 'gu', 'ar', '▁still', '▁travel', 's', '▁beyond', '▁the', '▁l', 'act', 'ed', '▁path', ',', '▁or', '▁the', '▁myth', 's', '▁of', '▁the', '▁ancient', '▁Inuit', '▁still', '▁resonate', ',', '▁or', '▁that', '▁in', '▁the', '▁Himalaya', 's', ',', '▁the', '▁Buddhist', 's', '▁continue', '▁to', '▁search', '▁for', '▁the', '▁breath', '▁of', '▁D', 'ha', 'r', 'ma', ',', '▁is', '▁to', '▁remember', '▁the', '▁basic', '▁insight', '▁of', '▁anthropology', ',', '▁and', '▁that', '▁means', '▁that', '▁the', '▁world', '▁in', '▁which', '▁we', '▁live', '▁does', 'n', "'", 't', '▁exist', '▁in', '▁an', '▁absolute', '▁sense', ',', '▁but', '▁is', '▁just', '▁an', '▁example', '▁of', '▁a', '▁consequence', '▁of', '▁a', '▁specific', '▁set', '▁of', '▁choices', '▁that', '▁our', '▁success', 'ion', '▁has', '▁several', '▁generations', '.'] 2024-01-16 15:16:40,263 - INFO - joeynmt.training - Example #1 2024-01-16 15:16:40,263 - INFO - joeynmt.training - Source: We're all born. We all bring our children into the world. 2024-01-16 15:16:40,263 - INFO - joeynmt.training - Reference: Nous sommes tous nés. Nous apportons nos enfants dans ce monde. 2024-01-16 15:16:40,264 - INFO - joeynmt.training - Hypothesis: Nous sommes tous nés. Nous apportons tous nos enfants dans le monde. 2024-01-16 15:16:40,264 - DEBUG - joeynmt.training - Tokenized source: ['▁We', "'", 're', '▁all', '▁born', '.', '▁We', '▁all', '▁bring', '▁our', '▁children', '▁into', '▁the', '▁world', '.'] 2024-01-16 15:16:40,264 - DEBUG - joeynmt.training - Tokenized reference: ['▁Nous', '▁sommes', '▁tous', '▁né', 's', '.', '▁Nous', '▁apport', 'ons', '▁nos', '▁enfants', '▁dans', '▁ce', '▁monde', '.'] 2024-01-16 15:16:40,264 - DEBUG - joeynmt.training - Tokenized hypothesis: ['', '▁Bien', '▁sûr', ',', '▁nous', '▁partageons', '▁tous', '▁les', '▁mêmes', '▁im', 'pé', 'r', 'atif', 's', '▁d', "'", 'adaptation', '.', '', '▁Nous', '▁sommes', '▁tous', '▁né', 's', '.', '▁Nous', '▁apport', 'ons', '▁tous', '▁nos', '▁enfants', '▁dans', '▁le', '▁monde', '.'] 2024-01-16 15:16:40,264 - INFO - joeynmt.training - Example #2 2024-01-16 15:16:40,265 - INFO - joeynmt.training - Source: Tous ces peuples nous enseignent qu'il y a d'autres façons d'être, d'autres façons de penser, d'autres manières de s'orienter sur Terre. 2024-01-16 15:16:40,265 - INFO - joeynmt.training - Reference: All of these peoples teach us that there are other ways of being, other ways of thinking, other ways of orienting yourself in the Earth. 2024-01-16 15:16:40,265 - INFO - joeynmt.training - Hypothesis: All these people teach us that there are other ways of being, other ways of thinking, other ways of orienting on Earth. 2024-01-16 15:16:40,265 - DEBUG - joeynmt.training - Tokenized source: ['▁Tous', '▁ces', '▁peuple', 's', '▁nous', '▁enseigne', 'nt', '▁qu', "'", 'il', '▁y', '▁a', '▁d', "'", 'autres', '▁façons', '▁d', "'", 'être', ',', '▁d', "'", 'autres', '▁façons', '▁de', '▁penser', ',', '▁d', "'", 'autres', '▁manières', '▁de', '▁s', "'", 'orient', 'er', '▁sur', '▁Terre', '.'] 2024-01-16 15:16:40,266 - DEBUG - joeynmt.training - Tokenized reference: ['▁All', '▁of', '▁these', '▁people', 's', '▁teach', '▁us', '▁that', '▁there', '▁are', '▁other', '▁ways', '▁of', '▁being', ',', '▁other', '▁ways', '▁of', '▁thinking', ',', '▁other', '▁ways', '▁of', '▁orient', 'ing', '▁yourself', '▁in', '▁the', '▁Earth', '.'] 2024-01-16 15:16:40,266 - DEBUG - joeynmt.training - Tokenized hypothesis: ['', '▁And', '▁whether', '▁it', '▁is', '▁the', '▁Pen', 'an', '▁in', '▁the', '▁forests', '▁of', '▁Bo', 'r', 'ne', 'o', ',', '▁or', '▁the', '▁V', 'oodoo', '▁a', 'co', 'ly', 'tes', '▁in', '▁Haiti', ',', '▁or', '▁the', '▁warrior', 's', '▁in', '▁the', '▁K', 'ais', 'ut', '▁desert', '▁of', '▁Northern', '▁Kenya', ',', '▁the', '▁C', 'ura', 'nder', 'o', '▁in', '▁the', '▁mountains', '▁of', '▁the', '▁And', 'es', ',', '▁or', '▁a', '▁car', 'ava', 'n', 's', 'er', 'ai', '▁in', '▁the', '▁middle', '▁of', '▁the', '▁Sahara', '▁--', '▁this', '▁is', '▁incidentally', '▁the', '▁fellow', '▁that', '▁I', '▁traveled', '▁into', '▁the', '▁desert', '▁with', '▁a', '▁month', '▁ago', '▁--', '▁or', '▁indeed', '▁a', '▁y', 'ak', '▁her', 'der', '▁in', '▁the', '▁slope', 's', '▁of', '▁Q', 'o', 'mo', 'lang', 'ma', ',', '▁Everest', ',', '▁the', '▁goddess', '▁mother', '▁of', '▁the', '▁world', '.', '', '▁All', '▁these', '▁people', '▁teach', '▁us', '▁that', '▁there', '▁are', '▁other', '▁ways', '▁of', '▁being', ',', '▁other', '▁ways', '▁of', '▁thinking', ',', '▁other', '▁ways', '▁of', '▁orient', 'ing', '▁on', '▁Earth', '.'] 2024-01-16 15:16:40,266 - INFO - joeynmt.training - Example #3 2024-01-16 15:16:40,266 - INFO - joeynmt.training - Source: And the great indicator of that, of course, is language loss. 2024-01-16 15:16:40,266 - INFO - joeynmt.training - Reference: Et l'indicateur le plus fiable est bien sûr l'extinction du langage. 2024-01-16 15:16:40,266 - INFO - joeynmt.training - Hypothesis: Et le grand indicateur de cela, bien sûr, est la perte de langue. 2024-01-16 15:16:40,267 - DEBUG - joeynmt.training - Tokenized source: ['▁And', '▁the', '▁great', '▁indicator', '▁of', '▁that', ',', '▁of', '▁course', ',', '▁is', '▁language', '▁loss', '.'] 2024-01-16 15:16:40,267 - DEBUG - joeynmt.training - Tokenized reference: ['▁Et', '▁l', "'", 'ind', 'ic', 'ateur', '▁le', '▁plus', '▁fiable', '▁est', '▁bien', '▁sûr', '▁l', "'", 'extinction', '▁du', '▁langage', '.'] 2024-01-16 15:16:40,267 - DEBUG - joeynmt.training - Tokenized hypothesis: ['', '▁Aucun', '▁biologist', 'e', ',', '▁par', '▁exemple', ',', '▁n', "'", 'ose', 'rait', '▁suggérer', '▁que', '▁50%', '▁ou', '▁plus', '▁de', '▁toutes', '▁les', '▁espèces', '▁ont', '▁été', '▁ou', '▁sont', '▁à', '▁deux', '▁doigts', '▁de', '▁l', "'", 'extinction', '▁parce', '▁que', '▁ce', '▁n', "'", 'est', '▁tout', '▁simplement', '▁pas', '▁vrai', ',', '▁et', '▁pourtant', '▁--', '▁que', '▁le', '▁scénario', '▁le', '▁plus', '▁a', 'po', 'ca', 'ly', 'p', 'tique', '▁dans', '▁le', '▁royaume', '▁de', '▁la', '▁diversité', '▁biologique', '▁--', '▁se', '▁rapproche', '▁rarement', '▁de', '▁ce', '▁que', '▁nous', '▁considér', 'ons', '▁comme', '▁le', '▁scénario', '▁le', '▁plus', '▁optimiste', '▁au', '▁sein', '▁de', '▁la', '▁diversité', '▁culturelle', '.', '', '▁Et', '▁le', '▁grand', '▁indicate', 'ur', '▁de', '▁cela', ',', '▁bien', '▁sûr', ',', '▁est', '▁la', '▁perte', '▁de', '▁langue', '.'] 2024-01-16 15:17:56,755 - INFO - joeynmt.training - Epoch 9, Step: 24100, Batch Loss: 1.051247, Batch Acc: 0.740390, Tokens per Sec: 16370, Lr: 0.000129 2024-01-16 15:19:13,500 - INFO - joeynmt.training - Epoch 9, Step: 24200, Batch Loss: 1.012516, Batch Acc: 0.740129, Tokens per Sec: 16332, Lr: 0.000129 2024-01-16 15:20:30,250 - INFO - joeynmt.training - Epoch 9, Step: 24300, Batch Loss: 1.046828, Batch Acc: 0.740927, Tokens per Sec: 16295, Lr: 0.000128 2024-01-16 15:21:46,285 - INFO - joeynmt.training - Epoch 9, Step: 24400, Batch Loss: 1.032326, Batch Acc: 0.740309, Tokens per Sec: 16460, Lr: 0.000128 2024-01-16 15:23:02,471 - INFO - joeynmt.training - Epoch 9, Step: 24500, Batch Loss: 1.045739, Batch Acc: 0.740893, Tokens per Sec: 16353, Lr: 0.000128 2024-01-16 15:24:17,797 - INFO - joeynmt.training - Epoch 9, Step: 24600, Batch Loss: 0.990458, Batch Acc: 0.741637, Tokens per Sec: 16511, Lr: 0.000128 2024-01-16 15:25:23,396 - INFO - joeynmt.training - Epoch 9, total training loss: 2837.93, num. of seqs: 702202, num. of tokens: 34266040, 2098.4420[sec] 2024-01-16 15:25:23,406 - INFO - joeynmt.training - EPOCH 10 2024-01-16 15:25:33,358 - INFO - joeynmt.training - Epoch 10, Step: 24700, Batch Loss: 0.941702, Batch Acc: 0.758506, Tokens per Sec: 16480, Lr: 0.000127 2024-01-16 15:26:50,944 - INFO - joeynmt.training - Epoch 10, Step: 24800, Batch Loss: 0.942496, Batch Acc: 0.757911, Tokens per Sec: 16107, Lr: 0.000127 2024-01-16 15:28:07,891 - INFO - joeynmt.training - Epoch 10, Step: 24900, Batch Loss: 1.000682, Batch Acc: 0.757235, Tokens per Sec: 16179, Lr: 0.000127 2024-01-16 15:29:24,131 - INFO - joeynmt.training - Epoch 10, Step: 25000, Batch Loss: 0.952470, Batch Acc: 0.755306, Tokens per Sec: 16422, Lr: 0.000126 2024-01-16 15:29:24,132 - INFO - joeynmt.datasets - Sample random subset from dev data: n=500, seed=25042 2024-01-16 15:29:24,132 - INFO - joeynmt.prediction - Predicting 500 example(s)... (Greedy decoding with min_output_length=1, max_output_length=512, return_prob='none', generate_unk=True, repetition_penalty=-1, no_repeat_ngram_size=-1) 2024-01-16 15:29:38,243 - INFO - joeynmt.prediction - Generation took 14.1023[sec]. 2024-01-16 15:29:38,377 - INFO - joeynmt.metrics - nrefs:1|case:lc|eff:no|tok:13a|smooth:exp|version:2.4.0 2024-01-16 15:29:38,377 - INFO - joeynmt.prediction - Evaluation result (greedy): bleu: 26.96, loss: 2.00, ppl: 7.42, acc: 0.59, 0.1164[sec] 2024-01-16 15:29:38,378 - INFO - joeynmt.training - Example #0 2024-01-16 15:29:38,379 - INFO - joeynmt.training - Source: Wade Davis sur les cultures en voie de disparition 2024-01-16 15:29:38,379 - INFO - joeynmt.training - Reference: Wade Davis: Dreams from endangered cultures 2024-01-16 15:29:38,379 - INFO - joeynmt.training - Hypothesis: Wade Davis on endangered crops. 2024-01-16 15:29:38,380 - DEBUG - joeynmt.training - Tokenized source: ['▁Wa', 'de', '▁Davis', '▁sur', '▁les', '▁cultures', '▁en', '▁voie', '▁de', '▁disparition'] 2024-01-16 15:29:38,380 - DEBUG - joeynmt.training - Tokenized reference: ['▁Wa', 'de', '▁Davis', ':', '▁Dream', 's', '▁from', '▁endangered', '▁cultures'] 2024-01-16 15:29:38,380 - DEBUG - joeynmt.training - Tokenized hypothesis: ['', '', '▁Wa', 'de', '▁Davis', '▁on', '▁endangered', '▁crops', '.'] 2024-01-16 15:29:38,380 - INFO - joeynmt.training - Example #1 2024-01-16 15:29:38,381 - INFO - joeynmt.training - Source: We go through initiation rites. 2024-01-16 15:29:38,381 - INFO - joeynmt.training - Reference: Nous procédons à des rites d'initiations. 2024-01-16 15:29:38,381 - INFO - joeynmt.training - Hypothesis: Nous passons par des rites d'initiation. 2024-01-16 15:29:38,381 - DEBUG - joeynmt.training - Tokenized source: ['▁We', '▁go', '▁through', '▁initiat', 'ion', '▁r', 'ites', '.'] 2024-01-16 15:29:38,381 - DEBUG - joeynmt.training - Tokenized reference: ['▁Nous', '▁procéd', 'ons', '▁à', '▁des', '▁r', 'ites', '▁d', "'", 'ini', 'ti', 'ations', '.'] 2024-01-16 15:29:38,382 - DEBUG - joeynmt.training - Tokenized hypothesis: ['', '▁Nous', '▁sommes', '▁tous', '▁né', 's', '.', '▁Nous', '▁apport', 'ons', '▁nos', '▁enfants', '▁dans', '▁ce', '▁monde', '.', '', '▁Nous', '▁passons', '▁par', '▁des', '▁r', 'ites', '▁d', "'", 'in', 'it', 'iation', '.'] 2024-01-16 15:29:38,382 - INFO - joeynmt.training - Example #2 2024-01-16 15:29:38,382 - INFO - joeynmt.training - Source: When each of you in this room were born, there were 6,000 languages spoken on the planet. 2024-01-16 15:29:38,382 - INFO - joeynmt.training - Reference: Lorsque chacun d'entre vous dans cette salle est né, 6,000 langues étaient pratiquées sur la planète. 2024-01-16 15:29:38,382 - INFO - joeynmt.training - Hypothesis: Quand chacun d'entre vous ici était né, il y avait 6000 langues parlé sur la planète. 2024-01-16 15:29:38,383 - DEBUG - joeynmt.training - Tokenized source: ['▁When', '▁each', '▁of', '▁you', '▁in', '▁this', '▁room', '▁were', '▁born', ',', '▁there', '▁were', '▁6,000', '▁languages', '▁spoken', '▁on', '▁the', '▁planet', '.'] 2024-01-16 15:29:38,383 - DEBUG - joeynmt.training - Tokenized reference: ['▁Lorsque', '▁chacun', '▁d', "'", 'entre', '▁vous', '▁dans', '▁cette', '▁salle', '▁est', '▁né', ',', '▁6,000', '▁langues', '▁étaient', '▁pratiqu', 'ées', '▁sur', '▁la', '▁planète', '.'] 2024-01-16 15:29:38,383 - DEBUG - joeynmt.training - Tokenized hypothesis: ['', '▁Et', '▁l', "'", 'ind', 'ic', 'ateur', '▁le', '▁plus', '▁fiable', '▁est', '▁bien', '▁sûr', '▁l', "'", 'extinction', '▁du', '▁langage', '.', '', '▁Quand', '▁chacun', '▁d', "'", 'entre', '▁vous', '▁ici', '▁était', '▁né', ',', '▁il', '▁y', '▁avait', '▁6000', '▁langues', '▁parlé', '▁sur', '▁la', '▁planète', '.'] 2024-01-16 15:29:38,383 - INFO - joeynmt.training - Example #3 2024-01-16 15:29:38,383 - INFO - joeynmt.training - Source: Et parmi ces 6,000 langues, alors que nous sommes à Monterey aujourd'hui, une bonne moitié n'est plus chuchotée dans les oreilles des enfants. 2024-01-16 15:29:38,383 - INFO - joeynmt.training - Reference: And of those 6,000 languages, as we sit here today in Monterey, fully half are no longer being whispered into the ears of children. 2024-01-16 15:29:38,383 - INFO - joeynmt.training - Hypothesis: And among those 6,000 languages, as we are in Monterey today, a good half is no longer in the ears of children. 2024-01-16 15:29:38,384 - DEBUG - joeynmt.training - Tokenized source: ['▁Et', '▁parmi', '▁ces', '▁6,000', '▁langues', ',', '▁alors', '▁que', '▁nous', '▁sommes', '▁à', '▁Monterey', '▁aujourd', "'", 'hui', ',', '▁une', '▁bonne', '▁moitié', '▁n', "'", 'est', '▁plus', '▁ch', 'uch', 'ot', 'ée', '▁dans', '▁les', '▁oreilles', '▁des', '▁enfants', '.'] 2024-01-16 15:29:38,384 - DEBUG - joeynmt.training - Tokenized reference: ['▁And', '▁of', '▁those', '▁6,000', '▁languages', ',', '▁as', '▁we', '▁sit', '▁here', '▁today', '▁in', '▁Monterey', ',', '▁fully', '▁half', '▁are', '▁no', '▁longer', '▁being', '▁whisper', 'ed', '▁into', '▁the', '▁ears', '▁of', '▁children', '.'] 2024-01-16 15:29:38,384 - DEBUG - joeynmt.training - Tokenized hypothesis: ['', '▁Every', '▁language', '▁is', '▁an', '▁old', '-', 'growth', '▁forest', '▁of', '▁the', '▁mind', ',', '▁a', '▁water', 'sh', 'ed', ',', '▁a', '▁thought', ',', '▁an', '▁ecosystem', '▁of', '▁spiritual', '▁possibilities', '.', '', '▁And', '▁among', '▁those', '▁6,000', '▁languages', ',', '▁as', '▁we', '▁are', '▁in', '▁Monterey', '▁today', ',', '▁a', '▁good', '▁half', '▁is', '▁no', '▁longer', '▁in', '▁the', '▁ears', '▁of', '▁children', '.'] 2024-01-16 15:30:54,488 - INFO - joeynmt.training - Epoch 10, Step: 25100, Batch Loss: 0.934536, Batch Acc: 0.756136, Tokens per Sec: 16442, Lr: 0.000126 2024-01-16 15:32:10,902 - INFO - joeynmt.training - Epoch 10, Step: 25200, Batch Loss: 0.971414, Batch Acc: 0.756177, Tokens per Sec: 16326, Lr: 0.000126 2024-01-16 15:33:27,833 - INFO - joeynmt.training - Epoch 10, Step: 25300, Batch Loss: 0.971244, Batch Acc: 0.755438, Tokens per Sec: 16311, Lr: 0.000126 2024-01-16 15:34:43,881 - INFO - joeynmt.training - Epoch 10, Step: 25400, Batch Loss: 1.001768, Batch Acc: 0.753701, Tokens per Sec: 16414, Lr: 0.000125 2024-01-16 15:36:00,381 - INFO - joeynmt.training - Epoch 10, Step: 25500, Batch Loss: 0.963367, Batch Acc: 0.754148, Tokens per Sec: 16423, Lr: 0.000125 2024-01-16 15:37:17,409 - INFO - joeynmt.training - Epoch 10, Step: 25600, Batch Loss: 0.963131, Batch Acc: 0.753832, Tokens per Sec: 16176, Lr: 0.000125 2024-01-16 15:38:34,030 - INFO - joeynmt.training - Epoch 10, Step: 25700, Batch Loss: 0.978197, Batch Acc: 0.754679, Tokens per Sec: 16302, Lr: 0.000125 2024-01-16 15:39:50,013 - INFO - joeynmt.training - Epoch 10, Step: 25800, Batch Loss: 0.957794, Batch Acc: 0.754520, Tokens per Sec: 16400, Lr: 0.000125 2024-01-16 15:41:06,282 - INFO - joeynmt.training - Epoch 10, Step: 25900, Batch Loss: 1.013107, Batch Acc: 0.754841, Tokens per Sec: 16406, Lr: 0.000124 2024-01-16 15:42:23,313 - INFO - joeynmt.training - Epoch 10, Step: 26000, Batch Loss: 0.992646, Batch Acc: 0.753480, Tokens per Sec: 16226, Lr: 0.000124 2024-01-16 15:42:23,314 - INFO - joeynmt.datasets - Sample random subset from dev data: n=500, seed=26042 2024-01-16 15:42:23,314 - INFO - joeynmt.prediction - Predicting 500 example(s)... (Greedy decoding with min_output_length=1, max_output_length=512, return_prob='none', generate_unk=True, repetition_penalty=-1, no_repeat_ngram_size=-1) 2024-01-16 15:42:40,915 - INFO - joeynmt.prediction - Generation took 17.5923[sec]. 2024-01-16 15:42:41,024 - INFO - joeynmt.metrics - nrefs:1|case:lc|eff:no|tok:13a|smooth:exp|version:2.4.0 2024-01-16 15:42:41,024 - INFO - joeynmt.prediction - Evaluation result (greedy): bleu: 27.30, loss: 2.05, ppl: 7.77, acc: 0.58, 0.0712[sec] 2024-01-16 15:42:41,025 - INFO - joeynmt.training - Example #0 2024-01-16 15:42:41,026 - INFO - joeynmt.training - Source: Just to know that Jaguar shamans still journey beyond the Milky Way, or the myths of the Inuit elders still resonate with meaning, or that in the Himalaya, the Buddhists still pursue the breath of the Dharma, is to really remember the central revelation of anthropology, and that is the idea that the world in which we live does not exist in some absolute sense, but is just one model of reality, the consequence of one particular set of adaptive choices that our lineage made, albeit successfully, many generations ago. 2024-01-16 15:42:41,026 - INFO - joeynmt.training - Reference: Le fait de savoir que les Jaguar shaman voyagent toujours au-delà de la voie lactée, ou que les mythes des anciens Inuit résonnent encore de sens, ou bien que dans l'Himalaya, les Bouddhistes continuent à rechercher le souffle du Dharma, c'est se rappeler de la révélation essentielle de l'anthropologie, et cela veut dire que le monde dans lequel nous vivons n'existe pas dans un sens absolu, mais est uniquement un exemple de réalité, la conséquence d'un ensemble spécifique de choix adaptés établis par notre lignée avec succès, il y a plusieurs générations. 2024-01-16 15:42:41,026 - INFO - joeynmt.training - Hypothesis: Juste pour savoir que les chamanes Jaguar continuent de se rendre au delà de la Voie Lactée, ou les mythes des anciens Inuits continuent à résonner dans le sens, ou dans l'Himalaya, les bouddhistes continuent à suivre la respiration du Dharme, se souviennent vraiment de la révélation centrale de l'anthropologie, et c'est l'idée que le monde dans lequel nous vivons n'existe pas dans un sens absolu, mais c'est juste un modèle de réalité, la conséquence d'un ensemble particulier de choix adaptatifs que notre lignée a fait, bien que nous avons réussi, il y a beaucoup de générations. 2024-01-16 15:42:41,028 - DEBUG - joeynmt.training - Tokenized source: ['▁Just', '▁to', '▁know', '▁that', '▁Ja', 'gu', 'ar', '▁shaman', 's', '▁still', '▁journey', '▁beyond', '▁the', '▁Milk', 'y', '▁Way', ',', '▁or', '▁the', '▁myth', 's', '▁of', '▁the', '▁Inuit', '▁el', 'der', 's', '▁still', '▁resonate', '▁with', '▁meaning', ',', '▁or', '▁that', '▁in', '▁the', '▁Himalaya', ',', '▁the', '▁Buddhist', 's', '▁still', '▁pursue', '▁the', '▁breath', '▁of', '▁the', '▁D', 'ha', 'r', 'ma', ',', '▁is', '▁to', '▁really', '▁remember', '▁the', '▁central', '▁revelation', '▁of', '▁anthropology', ',', '▁and', '▁that', '▁is', '▁the', '▁idea', '▁that', '▁the', '▁world', '▁in', '▁which', '▁we', '▁live', '▁does', '▁not', '▁exist', '▁in', '▁some', '▁absolute', '▁sense', ',', '▁but', '▁is', '▁just', '▁one', '▁model', '▁of', '▁reality', ',', '▁the', '▁consequence', '▁of', '▁one', '▁particular', '▁set', '▁of', '▁adaptive', '▁choices', '▁that', '▁our', '▁line', 'age', '▁made', ',', '▁albeit', '▁successfully', ',', '▁many', '▁generations', '▁ago', '.'] 2024-01-16 15:42:41,028 - DEBUG - joeynmt.training - Tokenized reference: ['▁Le', '▁fait', '▁de', '▁savoir', '▁que', '▁les', '▁Ja', 'gu', 'ar', '▁shaman', '▁voyage', 'nt', '▁toujours', '▁au', '-', 'delà', '▁de', '▁la', '▁voie', '▁l', 'act', 'ée', ',', '▁ou', '▁que', '▁les', '▁mythe', 's', '▁des', '▁anciens', '▁Inuit', '▁résonne', 'nt', '▁encore', '▁de', '▁sens', ',', '▁ou', '▁bien', '▁que', '▁dans', '▁l', "'", 'Himalaya', ',', '▁les', '▁Bo', 'ud', 'd', 'h', 'istes', '▁continuent', '▁à', '▁recherche', 'r', '▁le', '▁souffle', '▁du', '▁D', 'ha', 'r', 'ma', ',', '▁c', "'", 'est', '▁se', '▁rappeler', '▁de', '▁la', '▁révélation', '▁essentielle', '▁de', '▁l', "'", 'anthrop', 'ologie', ',', '▁et', '▁cela', '▁veut', '▁dire', '▁que', '▁le', '▁monde', '▁dans', '▁lequel', '▁nous', '▁vivons', '▁n', "'", 'existe', '▁pas', '▁dans', '▁un', '▁sens', '▁absolu', ',', '▁mais', '▁est', '▁uniquement', '▁un', '▁exemple', '▁de', '▁réalité', ',', '▁la', '▁conséquence', '▁d', "'", 'un', '▁ensemble', '▁spécifique', '▁de', '▁choix', '▁adapté', 's', '▁établi', 's', '▁par', '▁notre', '▁lignée', '▁avec', '▁succès', ',', '▁il', '▁y', '▁a', '▁plusieurs', '▁générations', '.'] 2024-01-16 15:42:41,028 - DEBUG - joeynmt.training - Tokenized hypothesis: ['', '▁Vous', '▁savez', ',', '▁un', '▁des', '▁plaisir', 's', '▁intense', 's', '▁du', '▁voyage', '▁et', '▁un', '▁des', '▁dé', 'lic', 'es', '▁de', '▁la', '▁recherche', '▁et', 'h', 'no', 'graph', 'ique', '▁est', '▁la', '▁possibilité', '▁de', '▁vivre', '▁parmi', '▁ceux', '▁qui', '▁n', "'", 'ont', '▁pas', '▁oublié', '▁les', '▁anciennes', '▁coutume', 's', ',', '▁qui', '▁ressentent', '▁encore', '▁leur', '▁passé', '▁souffle', 'r', '▁dans', '▁le', '▁vent', ',', '▁qui', '▁le', '▁touchent', '▁dans', '▁les', '▁pierres', '▁poli', 'es', '▁par', '▁la', '▁pluie', ',', '▁le', '▁dé', 'gu', 'sten', 't', '▁dans', '▁les', '▁feuilles', '▁a', 'mère', 's', '▁des', '▁plantes', '.', '', '▁Juste', '▁pour', '▁savoir', '▁que', '▁les', '▁cha', 'man', 'es', '▁Ja', 'gu', 'ar', '▁continuent', '▁de', '▁se', '▁rendre', '▁au', '▁de', 'là', '▁de', '▁la', '▁Voie', '▁Lactée', ',', '▁ou', '▁les', '▁mythe', 's', '▁des', '▁anciens', '▁Inuit', 's', '▁continuent', '▁à', '▁résonne', 'r', '▁dans', '▁le', '▁sens', ',', '▁ou', '▁dans', '▁l', "'", 'Himalaya', ',', '▁les', '▁bouddhiste', 's', '▁continuent', '▁à', '▁suivre', '▁la', '▁respiration', '▁du', '▁D', 'ha', 'r', 'me', ',', '▁se', '▁souviennent', '▁vraiment', '▁de', '▁la', '▁révélation', '▁centrale', '▁de', '▁l', "'", 'anthrop', 'ologie', ',', '▁et', '▁c', "'", 'est', '▁l', "'", 'idée', '▁que', '▁le', '▁monde', '▁dans', '▁lequel', '▁nous', '▁vivons', '▁n', "'", 'existe', '▁pas', '▁dans', '▁un', '▁sens', '▁absolu', ',', '▁mais', '▁c', "'", 'est', '▁juste', '▁un', '▁modèle', '▁de', '▁réalité', ',', '▁la', '▁conséquence', '▁d', "'", 'un', '▁ensemble', '▁particulier', '▁de', '▁choix', '▁adapt', 'atif', 's', '▁que', '▁notre', '▁lignée', '▁a', '▁fait', ',', '▁bien', '▁que', '▁nous', '▁avons', '▁réussi', ',', '▁il', '▁y', '▁a', '▁beaucoup', '▁de', '▁générations', '.'] 2024-01-16 15:42:41,028 - INFO - joeynmt.training - Example #1 2024-01-16 15:42:41,028 - INFO - joeynmt.training - Source: We're all born. We all bring our children into the world. 2024-01-16 15:42:41,028 - INFO - joeynmt.training - Reference: Nous sommes tous nés. Nous apportons nos enfants dans ce monde. 2024-01-16 15:42:41,028 - INFO - joeynmt.training - Hypothesis: Nous sommes tous nés. Nous apportons tous nos enfants dans le monde. 2024-01-16 15:42:41,029 - DEBUG - joeynmt.training - Tokenized source: ['▁We', "'", 're', '▁all', '▁born', '.', '▁We', '▁all', '▁bring', '▁our', '▁children', '▁into', '▁the', '▁world', '.'] 2024-01-16 15:42:41,029 - DEBUG - joeynmt.training - Tokenized reference: ['▁Nous', '▁sommes', '▁tous', '▁né', 's', '.', '▁Nous', '▁apport', 'ons', '▁nos', '▁enfants', '▁dans', '▁ce', '▁monde', '.'] 2024-01-16 15:42:41,029 - DEBUG - joeynmt.training - Tokenized hypothesis: ['', '▁Bien', '▁sûr', ',', '▁nous', '▁partageons', '▁tous', '▁les', '▁mêmes', '▁im', 'pé', 'r', 'atif', 's', '▁d', "'", 'adaptation', '.', '', '▁Nous', '▁sommes', '▁tous', '▁né', 's', '.', '▁Nous', '▁apport', 'ons', '▁tous', '▁nos', '▁enfants', '▁dans', '▁le', '▁monde', '.'] 2024-01-16 15:42:41,029 - INFO - joeynmt.training - Example #2 2024-01-16 15:42:41,030 - INFO - joeynmt.training - Source: Tous ces peuples nous enseignent qu'il y a d'autres façons d'être, d'autres façons de penser, d'autres manières de s'orienter sur Terre. 2024-01-16 15:42:41,030 - INFO - joeynmt.training - Reference: All of these peoples teach us that there are other ways of being, other ways of thinking, other ways of orienting yourself in the Earth. 2024-01-16 15:42:41,030 - INFO - joeynmt.training - Hypothesis: All these people teach us that there are other ways of being, other ways of thinking, other ways of orienting on Earth. 2024-01-16 15:42:41,030 - DEBUG - joeynmt.training - Tokenized source: ['▁Tous', '▁ces', '▁peuple', 's', '▁nous', '▁enseigne', 'nt', '▁qu', "'", 'il', '▁y', '▁a', '▁d', "'", 'autres', '▁façons', '▁d', "'", 'être', ',', '▁d', "'", 'autres', '▁façons', '▁de', '▁penser', ',', '▁d', "'", 'autres', '▁manières', '▁de', '▁s', "'", 'orient', 'er', '▁sur', '▁Terre', '.'] 2024-01-16 15:42:41,031 - DEBUG - joeynmt.training - Tokenized reference: ['▁All', '▁of', '▁these', '▁people', 's', '▁teach', '▁us', '▁that', '▁there', '▁are', '▁other', '▁ways', '▁of', '▁being', ',', '▁other', '▁ways', '▁of', '▁thinking', ',', '▁other', '▁ways', '▁of', '▁orient', 'ing', '▁yourself', '▁in', '▁the', '▁Earth', '.'] 2024-01-16 15:42:41,031 - DEBUG - joeynmt.training - Tokenized hypothesis: ['', '▁And', '▁whether', '▁it', '▁is', '▁the', '▁Pen', 'an', '▁in', '▁the', '▁forests', '▁of', '▁Bo', 'r', 'ne', 'o', ',', '▁or', '▁the', '▁V', 'oodoo', '▁a', 'co', 'ly', 'tes', '▁in', '▁Haiti', ',', '▁or', '▁the', '▁warrior', 's', '▁in', '▁the', '▁K', 'ais', 'ut', '▁desert', '▁of', '▁Northern', '▁Kenya', ',', '▁the', '▁C', 'ura', 'nder', 'o', '▁in', '▁the', '▁mountains', '▁of', '▁the', '▁And', 'es', ',', '▁or', '▁a', '▁car', 'ava', 'n', 's', 'er', 'ai', '▁in', '▁the', '▁middle', '▁of', '▁the', '▁Sahara', '▁--', '▁this', '▁is', '▁incidentally', '▁the', '▁fellow', '▁that', '▁I', '▁traveled', '▁into', '▁the', '▁desert', '▁with', '▁a', '▁month', '▁ago', '▁--', '▁or', '▁indeed', '▁a', '▁y', 'ak', '▁her', 'der', '▁in', '▁the', '▁slope', 's', '▁of', '▁Q', 'o', 'mo', 'lang', 'ma', ',', '▁Everest', ',', '▁the', '▁goddess', '▁mother', '▁of', '▁the', '▁world', '.', '', '▁All', '▁these', '▁people', '▁teach', '▁us', '▁that', '▁there', '▁are', '▁other', '▁ways', '▁of', '▁being', ',', '▁other', '▁ways', '▁of', '▁thinking', ',', '▁other', '▁ways', '▁of', '▁orient', 'ing', '▁on', '▁Earth', '.'] 2024-01-16 15:42:41,031 - INFO - joeynmt.training - Example #3 2024-01-16 15:42:41,031 - INFO - joeynmt.training - Source: Now, together the myriad cultures of the world make up a web of spiritual life and cultural life that envelops the planet, and is as important to the well-being of the planet as indeed is the biological web of life that you know as a biosphere. 2024-01-16 15:42:41,031 - INFO - joeynmt.training - Reference: Aujourd'hui, les innombrables cultures dans le monde constituent un tissu de vie spirituelle et culturelle qui enveloppe la planète, et qui est aussi important pour le bien-être de la planète que l'est également le tissu biologique de la vie que vous connaissez en tant que biosphère. 2024-01-16 15:42:41,031 - INFO - joeynmt.training - Hypothesis: Maintenant, ensemble, les myriades de cultures du monde constituent un web de vie spirituelle et de vie culturelle qui enveloppent la planète, et qui est aussi important pour le bien-être de la planète que le web biologique de la vie que vous connaissez en tant que biosphère. 2024-01-16 15:42:41,032 - DEBUG - joeynmt.training - Tokenized source: ['▁Now', ',', '▁together', '▁the', '▁myriad', '▁cultures', '▁of', '▁the', '▁world', '▁make', '▁up', '▁a', '▁web', '▁of', '▁spiritual', '▁life', '▁and', '▁cultural', '▁life', '▁that', '▁en', 've', 'lop', 's', '▁the', '▁planet', ',', '▁and', '▁is', '▁as', '▁important', '▁to', '▁the', '▁well', '-', 'being', '▁of', '▁the', '▁planet', '▁as', '▁indeed', '▁is', '▁the', '▁biological', '▁web', '▁of', '▁life', '▁that', '▁you', '▁know', '▁as', '▁a', '▁biosphere', '.'] 2024-01-16 15:42:41,032 - DEBUG - joeynmt.training - Tokenized reference: ['▁A', 'ujourd', "'", 'hui', ',', '▁les', '▁', 'innombrables', '▁cultures', '▁dans', '▁le', '▁monde', '▁constituent', '▁un', '▁tissu', '▁de', '▁vie', '▁spirituelle', '▁et', '▁culturelle', '▁qui', '▁enveloppe', '▁la', '▁planète', ',', '▁et', '▁qui', '▁est', '▁aussi', '▁important', '▁pour', '▁le', '▁bien', '-', 'être', '▁de', '▁la', '▁planète', '▁que', '▁l', "'", 'est', '▁également', '▁le', '▁tissu', '▁biologique', '▁de', '▁la', '▁vie', '▁que', '▁vous', '▁connaissez', '▁en', '▁tant', '▁que', '▁biosphère', '.'] 2024-01-16 15:42:41,032 - DEBUG - joeynmt.training - Tokenized hypothesis: ['', '▁Et', '▁ceci', '▁est', '▁une', '▁idée', ',', '▁si', '▁on', '▁y', '▁réfléchi', 't', ',', '▁qui', '▁ne', '▁peut', '▁que', '▁vous', '▁remplir', '▁d', "'", 'espoir', '.', '', '▁Maintenant', ',', '▁ensemble', ',', '▁les', '▁myriad', 'es', '▁de', '▁cultures', '▁du', '▁monde', '▁constituent', '▁un', '▁web', '▁de', '▁vie', '▁spirituelle', '▁et', '▁de', '▁vie', '▁culturelle', '▁qui', '▁en', 've', 'lop', 'p', 'ent', '▁la', '▁planète', ',', '▁et', '▁qui', '▁est', '▁aussi', '▁important', '▁pour', '▁le', '▁bien', '-', 'être', '▁de', '▁la', '▁planète', '▁que', '▁le', '▁web', '▁biologique', '▁de', '▁la', '▁vie', '▁que', '▁vous', '▁connaissez', '▁en', '▁tant', '▁que', '▁biosphère', '.'] 2024-01-16 15:43:57,181 - INFO - joeynmt.training - Epoch 10, Step: 26100, Batch Loss: 0.953479, Batch Acc: 0.753515, Tokens per Sec: 16385, Lr: 0.000124 2024-01-16 15:45:13,764 - INFO - joeynmt.training - Epoch 10, Step: 26200, Batch Loss: 0.926838, Batch Acc: 0.753715, Tokens per Sec: 16338, Lr: 0.000124 2024-01-16 15:46:29,534 - INFO - joeynmt.training - Epoch 10, Step: 26300, Batch Loss: 0.948857, Batch Acc: 0.752772, Tokens per Sec: 16386, Lr: 0.000123 2024-01-16 15:47:46,290 - INFO - joeynmt.training - Epoch 10, Step: 26400, Batch Loss: 1.013003, Batch Acc: 0.753225, Tokens per Sec: 16254, Lr: 0.000123 2024-01-16 15:49:03,141 - INFO - joeynmt.training - Epoch 10, Step: 26500, Batch Loss: 0.970040, Batch Acc: 0.754576, Tokens per Sec: 16309, Lr: 0.000123 2024-01-16 15:50:18,947 - INFO - joeynmt.training - Epoch 10, Step: 26600, Batch Loss: 0.987390, Batch Acc: 0.753705, Tokens per Sec: 16539, Lr: 0.000123 2024-01-16 15:51:35,692 - INFO - joeynmt.training - Epoch 10, Step: 26700, Batch Loss: 0.971609, Batch Acc: 0.753316, Tokens per Sec: 16305, Lr: 0.000122 2024-01-16 15:52:51,754 - INFO - joeynmt.training - Epoch 10, Step: 26800, Batch Loss: 1.018167, Batch Acc: 0.752278, Tokens per Sec: 16472, Lr: 0.000122 2024-01-16 15:54:08,010 - INFO - joeynmt.training - Epoch 10, Step: 26900, Batch Loss: 0.984442, Batch Acc: 0.752959, Tokens per Sec: 16396, Lr: 0.000122 2024-01-16 15:55:24,205 - INFO - joeynmt.training - Epoch 10, Step: 27000, Batch Loss: 0.992056, Batch Acc: 0.752779, Tokens per Sec: 16344, Lr: 0.000122 2024-01-16 15:55:24,206 - INFO - joeynmt.datasets - Sample random subset from dev data: n=500, seed=27042 2024-01-16 15:55:24,206 - INFO - joeynmt.prediction - Predicting 500 example(s)... (Greedy decoding with min_output_length=1, max_output_length=512, return_prob='none', generate_unk=True, repetition_penalty=-1, no_repeat_ngram_size=-1) 2024-01-16 15:55:36,289 - INFO - joeynmt.prediction - Generation took 12.0745[sec]. 2024-01-16 15:55:36,374 - INFO - joeynmt.metrics - nrefs:1|case:lc|eff:no|tok:13a|smooth:exp|version:2.4.0 2024-01-16 15:55:36,374 - INFO - joeynmt.prediction - Evaluation result (greedy): bleu: 28.05, loss: 1.95, ppl: 7.02, acc: 0.60, 0.0683[sec] 2024-01-16 15:55:39,254 - INFO - joeynmt.training - Checkpoint saved in models/iwslt14_prompt/27000.ckpt. 2024-01-16 15:55:39,273 - INFO - joeynmt.training - Example #0 2024-01-16 15:55:39,274 - INFO - joeynmt.training - Source: Avec des photos et des histoires passionnantes, l'explorateur du National Geographic Wade Davis rend hommage à l'extraordinaire diversité des cultures indigènes du monde, qui sont en voie de disparition sur notre planète à une vitesse alarmante. 2024-01-16 15:55:39,274 - INFO - joeynmt.training - Reference: With stunning photos and stories, National Geographic Explorer Wade Davis celebrates the extraordinary diversity of the world's indigenous cultures, which are disappearing from the planet at an alarming rate. 2024-01-16 15:55:39,274 - INFO - joeynmt.training - Hypothesis: With exciting pictures and stories, the National Geographic Filmboardexplorator is taking tribute to the extraordinary diversity of indigenous cultures around the world, which are disappearing on our planet at a alarming speed. 2024-01-16 15:55:39,275 - DEBUG - joeynmt.training - Tokenized source: ['▁Avec', '▁des', '▁photos', '▁et', '▁des', '▁histoires', '▁passionnant', 'es', ',', '▁l', "'", 'exp', 'lo', 'r', 'ateur', '▁du', '▁National', '▁Geographic', '▁Wa', 'de', '▁Davis', '▁rend', '▁hommage', '▁à', '▁l', "'", 'extraordinaire', '▁diversité', '▁des', '▁cultures', '▁in', 'dig', 'ène', 's', '▁du', '▁monde', ',', '▁qui', '▁sont', '▁en', '▁voie', '▁de', '▁disparition', '▁sur', '▁notre', '▁planète', '▁à', '▁une', '▁vitesse', '▁alarm', 'ante', '.'] 2024-01-16 15:55:39,275 - DEBUG - joeynmt.training - Tokenized reference: ['▁With', '▁stunning', '▁photos', '▁and', '▁stories', ',', '▁National', '▁Geographic', '▁Explorer', '▁Wa', 'de', '▁Davis', '▁celebrate', 's', '▁the', '▁extraordinary', '▁diversity', '▁of', '▁the', '▁world', "'", 's', '▁indige', 'nous', '▁cultures', ',', '▁which', '▁are', '▁disappearing', '▁from', '▁the', '▁planet', '▁at', '▁an', '▁alarm', 'ing', '▁rate', '.'] 2024-01-16 15:55:39,275 - DEBUG - joeynmt.training - Tokenized hypothesis: ['', '', '▁With', '▁exciting', '▁pictures', '▁and', '▁stories', ',', '▁the', '▁National', '▁Geographic', '▁Film', 'board', 'exp', 'lo', 'r', 'ator', '▁is', '▁taking', '▁tribute', '▁to', '▁the', '▁extraordinary', '▁diversity', '▁of', '▁indige', 'nous', '▁cultures', '▁around', '▁the', '▁world', ',', '▁which', '▁are', '▁disappearing', '▁on', '▁our', '▁planet', '▁at', '▁a', '▁alarm', 'ing', '▁speed', '.'] 2024-01-16 15:55:39,275 - INFO - joeynmt.training - Example #1 2024-01-16 15:55:39,275 - INFO - joeynmt.training - Source: Bien sûr, nous partageons tous les mêmes impératifs d'adaptation. 2024-01-16 15:55:39,275 - INFO - joeynmt.training - Reference: And of course, we all share the same adaptive imperatives. 2024-01-16 15:55:39,275 - INFO - joeynmt.training - Hypothesis: Of course, we all share the same imperatives of adaptation. 2024-01-16 15:55:39,276 - DEBUG - joeynmt.training - Tokenized source: ['▁Bien', '▁sûr', ',', '▁nous', '▁partageons', '▁tous', '▁les', '▁mêmes', '▁im', 'pé', 'r', 'atif', 's', '▁d', "'", 'adaptation', '.'] 2024-01-16 15:55:39,276 - DEBUG - joeynmt.training - Tokenized reference: ['▁And', '▁of', '▁course', ',', '▁we', '▁all', '▁share', '▁the', '▁same', '▁adaptive', '▁imperative', 's', '.'] 2024-01-16 15:55:39,276 - DEBUG - joeynmt.training - Tokenized hypothesis: ['', '▁Just', '▁to', '▁know', '▁that', '▁Ja', 'gu', 'ar', '▁shaman', 's', '▁still', '▁journey', '▁beyond', '▁the', '▁Milk', 'y', '▁Way', ',', '▁or', '▁the', '▁myth', 's', '▁of', '▁the', '▁Inuit', '▁el', 'der', 's', '▁still', '▁resonate', '▁with', '▁meaning', ',', '▁or', '▁that', '▁in', '▁the', '▁Himalaya', ',', '▁the', '▁Buddhist', 's', '▁still', '▁pursue', '▁the', '▁breath', '▁of', '▁the', '▁D', 'ha', 'r', 'ma', ',', '▁is', '▁to', '▁really', '▁remember', '▁the', '▁central', '▁revelation', '▁of', '▁anthropology', ',', '▁and', '▁that', '▁is', '▁the', '▁idea', '▁that', '▁the', '▁world', '▁in', '▁which', '▁we', '▁live', '▁does', '▁not', '▁exist', '▁in', '▁some', '▁absolute', '▁sense', ',', '▁but', '▁is', '▁just', '▁one', '▁model', '▁of', '▁reality', ',', '▁the', '▁consequence', '▁of', '▁one', '▁particular', '▁set', '▁of', '▁adaptive', '▁choices', '▁that', '▁our', '▁line', 'age', '▁made', ',', '▁albeit', '▁successfully', ',', '▁many', '▁generations', '▁ago', '.', '', '▁Of', '▁course', ',', '▁we', '▁all', '▁share', '▁the', '▁same', '▁imperative', 's', '▁of', '▁adaptation', '.'] 2024-01-16 15:55:39,276 - INFO - joeynmt.training - Example #2 2024-01-16 15:55:39,277 - INFO - joeynmt.training - Source: Et lorsque la biosphère fut sérieusement érodée, l'ethnosphère l'a été également -- et peut-être bien plus rapidement. 2024-01-16 15:55:39,277 - INFO - joeynmt.training - Reference: And just as the biosphere has been severely eroded, so too is the ethnosphere -- and, if anything, at a far greater rate. 2024-01-16 15:55:39,277 - INFO - joeynmt.training - Hypothesis: And when the biosphere was seriously degraded, the athnospher was also -- and perhaps much more quickly. 2024-01-16 15:55:39,277 - DEBUG - joeynmt.training - Tokenized source: ['▁Et', '▁lorsque', '▁la', '▁biosphère', '▁fut', '▁sérieusement', '▁é', 'ro', 'd', 'ée', ',', '▁l', "'", 'e', 'th', 'no', 's', 'ph', 'ère', '▁l', "'", 'a', '▁été', '▁également', '▁--', '▁et', '▁peut', '-', 'être', '▁bien', '▁plus', '▁rapidement', '.'] 2024-01-16 15:55:39,278 - DEBUG - joeynmt.training - Tokenized reference: ['▁And', '▁just', '▁as', '▁the', '▁biosphere', '▁has', '▁been', '▁severely', '▁er', 'od', 'ed', ',', '▁so', '▁too', '▁is', '▁the', '▁et', 'h', 'n', 'osphere', '▁--', '▁and', ',', '▁if', '▁anything', ',', '▁at', '▁a', '▁far', '▁greater', '▁rate', '.'] 2024-01-16 15:55:39,278 - DEBUG - joeynmt.training - Tokenized hypothesis: ['', '▁It', "'", 's', '▁the', '▁symbol', '▁of', '▁all', '▁that', '▁we', '▁are', '▁and', '▁all', '▁that', '▁we', '▁can', '▁be', '▁as', '▁an', '▁astonishing', 'ly', '▁in', 'qui', 's', 'it', 'ive', '▁species', '.', '', '▁And', '▁when', '▁the', '▁biosphere', '▁was', '▁seriously', '▁degraded', ',', '▁the', '▁a', 'th', 'no', 's', 'ph', 'er', '▁was', '▁also', '▁--', '▁and', '▁perhaps', '▁much', '▁more', '▁quickly', '.'] 2024-01-16 15:55:39,278 - INFO - joeynmt.training - Example #3 2024-01-16 15:55:39,278 - INFO - joeynmt.training - Source: They're no longer being taught to babies, which means, effectively, unless something changes, they're already dead. 2024-01-16 15:55:39,278 - INFO - joeynmt.training - Reference: Elles ne sont plus enseignées aux bébés, ce qui veut effectivement dire qu'à moins qu'un changement ait lieu, elles sont déjà mortes. 2024-01-16 15:55:39,278 - INFO - joeynmt.training - Hypothesis: Ils ne sont plus enseignés aux bébés, ce qui veut dire, effectivement, à moins que quelque chose ne change, ils sont déjà morts. 2024-01-16 15:55:39,279 - DEBUG - joeynmt.training - Tokenized source: ['▁They', "'", 're', '▁no', '▁longer', '▁being', '▁taught', '▁to', '▁babies', ',', '▁which', '▁means', ',', '▁effectively', ',', '▁unless', '▁something', '▁changes', ',', '▁they', "'", 're', '▁already', '▁dead', '.'] 2024-01-16 15:55:39,279 - DEBUG - joeynmt.training - Tokenized reference: ['▁Elles', '▁ne', '▁sont', '▁plus', '▁enseigné', 'es', '▁aux', '▁bébés', ',', '▁ce', '▁qui', '▁veut', '▁effectivement', '▁dire', '▁qu', "'", 'à', '▁moins', '▁qu', "'", 'un', '▁changement', '▁ait', '▁lieu', ',', '▁elles', '▁sont', '▁déjà', '▁morte', 's', '.'] 2024-01-16 15:55:39,279 - DEBUG - joeynmt.training - Tokenized hypothesis: ['', '▁Et', '▁parmi', '▁ces', '▁6,000', '▁langues', ',', '▁alors', '▁que', '▁nous', '▁sommes', '▁à', '▁Monterey', '▁aujourd', "'", 'hui', ',', '▁une', '▁bonne', '▁moitié', '▁n', "'", 'est', '▁plus', '▁ch', 'uch', 'ot', 'ée', '▁dans', '▁les', '▁oreilles', '▁des', '▁enfants', '.', '', '▁Ils', '▁ne', '▁sont', '▁plus', '▁enseigné', 's', '▁aux', '▁bébés', ',', '▁ce', '▁qui', '▁veut', '▁dire', ',', '▁effectivement', ',', '▁à', '▁moins', '▁que', '▁quelque', '▁chose', '▁ne', '▁change', ',', '▁ils', '▁sont', '▁déjà', '▁morts', '.'] 2024-01-16 15:56:54,409 - INFO - joeynmt.training - Epoch 10, Step: 27100, Batch Loss: 0.991460, Batch Acc: 0.753488, Tokens per Sec: 16599, Lr: 0.000121 2024-01-16 15:58:11,222 - INFO - joeynmt.training - Epoch 10, Step: 27200, Batch Loss: 0.966526, Batch Acc: 0.752444, Tokens per Sec: 16346, Lr: 0.000121 2024-01-16 15:59:27,342 - INFO - joeynmt.training - Epoch 10, Step: 27300, Batch Loss: 0.948346, Batch Acc: 0.752615, Tokens per Sec: 16337, Lr: 0.000121 2024-01-16 16:00:43,340 - INFO - joeynmt.training - Epoch 10, Step: 27400, Batch Loss: 0.974254, Batch Acc: 0.753144, Tokens per Sec: 16320, Lr: 0.000121 2024-01-16 16:01:06,495 - INFO - joeynmt.training - Epoch 10, total training loss: 2666.34, num. of seqs: 702202, num. of tokens: 34266040, 2096.0144[sec] 2024-01-16 16:01:06,505 - INFO - joeynmt.training - EPOCH 11 2024-01-16 16:02:00,017 - INFO - joeynmt.training - Epoch 11, Step: 27500, Batch Loss: 0.917497, Batch Acc: 0.770796, Tokens per Sec: 16304, Lr: 0.000121 2024-01-16 16:03:16,487 - INFO - joeynmt.training - Epoch 11, Step: 27600, Batch Loss: 0.927255, Batch Acc: 0.769529, Tokens per Sec: 16303, Lr: 0.000120 2024-01-16 16:04:32,321 - INFO - joeynmt.training - Epoch 11, Step: 27700, Batch Loss: 0.886569, Batch Acc: 0.769062, Tokens per Sec: 16599, Lr: 0.000120 2024-01-16 16:05:47,685 - INFO - joeynmt.training - Epoch 11, Step: 27800, Batch Loss: 0.927977, Batch Acc: 0.768535, Tokens per Sec: 16529, Lr: 0.000120 2024-01-16 16:07:03,822 - INFO - joeynmt.training - Epoch 11, Step: 27900, Batch Loss: 0.928414, Batch Acc: 0.766583, Tokens per Sec: 16435, Lr: 0.000120 2024-01-16 16:08:20,170 - INFO - joeynmt.training - Epoch 11, Step: 28000, Batch Loss: 0.912511, Batch Acc: 0.767491, Tokens per Sec: 16401, Lr: 0.000120 2024-01-16 16:08:20,171 - INFO - joeynmt.datasets - Sample random subset from dev data: n=500, seed=28042 2024-01-16 16:08:20,171 - INFO - joeynmt.prediction - Predicting 500 example(s)... (Greedy decoding with min_output_length=1, max_output_length=512, return_prob='none', generate_unk=True, repetition_penalty=-1, no_repeat_ngram_size=-1) 2024-01-16 16:08:47,254 - INFO - joeynmt.prediction - Generation took 27.0740[sec]. 2024-01-16 16:08:47,387 - INFO - joeynmt.metrics - nrefs:1|case:lc|eff:no|tok:13a|smooth:exp|version:2.4.0 2024-01-16 16:08:47,387 - INFO - joeynmt.prediction - Evaluation result (greedy): bleu: 26.19, loss: 2.05, ppl: 7.74, acc: 0.59, 0.1155[sec] 2024-01-16 16:08:47,388 - INFO - joeynmt.training - Example #0 2024-01-16 16:08:47,389 - INFO - joeynmt.training - Source: Par contre, la cadence exceptionnelle de la chanson est intéressante, le rythme de la danse dans toutes les cultures. 2024-01-16 16:08:47,389 - INFO - joeynmt.training - Reference: But what's interesting is the unique cadence of the song, the rhythm of the dance in every culture. 2024-01-16 16:08:47,389 - INFO - joeynmt.training - Hypothesis: Instead, the exceptional cadence of the song is interesting, the rhythm of dance in all cultures. 2024-01-16 16:08:47,390 - DEBUG - joeynmt.training - Tokenized source: ['▁Par', '▁contre', ',', '▁la', '▁cadence', '▁exceptionnel', 'le', '▁de', '▁la', '▁chanson', '▁est', '▁intéressante', ',', '▁le', '▁rythme', '▁de', '▁la', '▁dans', 'e', '▁dans', '▁toutes', '▁les', '▁cultures', '.'] 2024-01-16 16:08:47,390 - DEBUG - joeynmt.training - Tokenized reference: ['▁But', '▁what', "'", 's', '▁interesting', '▁is', '▁the', '▁unique', '▁cadence', '▁of', '▁the', '▁song', ',', '▁the', '▁rhythm', '▁of', '▁the', '▁dance', '▁in', '▁every', '▁culture', '.'] 2024-01-16 16:08:47,390 - DEBUG - joeynmt.training - Tokenized hypothesis: ['', '▁We', '▁have', '▁to', '▁deal', '▁with', '▁the', '▁inexorable', '▁separation', '▁of', '▁death', ',', '▁so', '▁it', '▁should', 'n', "'", 't', '▁surprise', '▁us', '▁that', '▁we', '▁all', '▁sing', ',', '▁we', '▁all', '▁dance', ',', '▁we', '▁all', '▁have', '▁art', '.', '', '▁Instead', ',', '▁the', '▁exceptional', '▁cadence', '▁of', '▁the', '▁song', '▁is', '▁interesting', ',', '▁the', '▁rhythm', '▁of', '▁dance', '▁in', '▁all', '▁cultures', '.'] 2024-01-16 16:08:47,390 - INFO - joeynmt.training - Example #1 2024-01-16 16:08:47,391 - INFO - joeynmt.training - Source: Et que ce soit le Penan dans les forêts du Bornéo, ou les acolytes Voodoo à Haïti, ou bien les guerriers dans le désert du Kaisut au nord du Kenya, le Curendero dans les montagnes des Andes, ou bien un caravansérail en plein milieu du Sahara. A propos, c'est la personne avec qui j'ai voyagé dans le désert il y un mois, ou effectivement, le gardien de troupeau de Yaks sur les flancs du Qomolangma, l'Everest, la déesse du monde. 2024-01-16 16:08:47,391 - INFO - joeynmt.training - Reference: And whether it is the Penan in the forests of Borneo, or the Voodoo acolytes in Haiti, or the warriors in the Kaisut desert of Northern Kenya, the Curandero in the mountains of the Andes, or a caravanserai in the middle of the Sahara -- this is incidentally the fellow that I traveled into the desert with a month ago -- or indeed a yak herder in the slopes of Qomolangma, Everest, the goddess mother of the world. 2024-01-16 16:08:47,391 - INFO - joeynmt.training - Hypothesis: And whether it's the Penan in the forests of Borneo, or the Voodoo Acolytes in Haiti, or the warriors in the northern Kenyan desert, the Curendero in the mountains of the Andes, or a caravanshe in the middle of the Sahara. And by the way, this is the person I traveled to the desert a month ago, or indeed, the Yak's herds on the flancs of Qomolangma, Mt. Everest, the global goddess. 2024-01-16 16:08:47,392 - DEBUG - joeynmt.training - Tokenized source: ['▁Et', '▁que', '▁ce', '▁soit', '▁le', '▁Pen', 'an', '▁dans', '▁les', '▁forêts', '▁du', '▁Bo', 'r', 'né', 'o', ',', '▁ou', '▁les', '▁a', 'co', 'ly', 'tes', '▁V', 'oodoo', '▁à', '▁Haïti', ',', '▁ou', '▁bien', '▁les', '▁guerrier', 's', '▁dans', '▁le', '▁désert', '▁du', '▁K', 'ais', 'ut', '▁au', '▁nord', '▁du', '▁Kenya', ',', '▁le', '▁C', 'ur', 'ende', 'ro', '▁dans', '▁les', '▁montagnes', '▁des', '▁And', 'es', ',', '▁ou', '▁bien', '▁un', '▁car', 'ava', 'n', 's', 'é', 'ra', 'il', '▁en', '▁plein', '▁milieu', '▁du', '▁Sahara', '.', '▁A', '▁propos', ',', '▁c', "'", 'est', '▁la', '▁personne', '▁avec', '▁qui', '▁j', "'", 'ai', '▁voyagé', '▁dans', '▁le', '▁désert', '▁il', '▁y', '▁un', '▁mois', ',', '▁ou', '▁effectivement', ',', '▁le', '▁gardien', '▁de', '▁troupeau', '▁de', '▁Yak', 's', '▁sur', '▁les', '▁fla', 'n', 'c', 's', '▁du', '▁Q', 'o', 'mo', 'lang', 'ma', ',', '▁l', "'", 'Everest', ',', '▁la', '▁dé', 'esse', '▁du', '▁monde', '.'] 2024-01-16 16:08:47,392 - DEBUG - joeynmt.training - Tokenized reference: ['▁And', '▁whether', '▁it', '▁is', '▁the', '▁Pen', 'an', '▁in', '▁the', '▁forests', '▁of', '▁Bo', 'r', 'ne', 'o', ',', '▁or', '▁the', '▁V', 'oodoo', '▁a', 'co', 'ly', 'tes', '▁in', '▁Haiti', ',', '▁or', '▁the', '▁warrior', 's', '▁in', '▁the', '▁K', 'ais', 'ut', '▁desert', '▁of', '▁Northern', '▁Kenya', ',', '▁the', '▁C', 'ura', 'nder', 'o', '▁in', '▁the', '▁mountains', '▁of', '▁the', '▁And', 'es', ',', '▁or', '▁a', '▁car', 'ava', 'n', 's', 'er', 'ai', '▁in', '▁the', '▁middle', '▁of', '▁the', '▁Sahara', '▁--', '▁this', '▁is', '▁incidentally', '▁the', '▁fellow', '▁that', '▁I', '▁traveled', '▁into', '▁the', '▁desert', '▁with', '▁a', '▁month', '▁ago', '▁--', '▁or', '▁indeed', '▁a', '▁y', 'ak', '▁her', 'der', '▁in', '▁the', '▁slope', 's', '▁of', '▁Q', 'o', 'mo', 'lang', 'ma', ',', '▁Everest', ',', '▁the', '▁goddess', '▁mother', '▁of', '▁the', '▁world', '.'] 2024-01-16 16:08:47,392 - DEBUG - joeynmt.training - Tokenized hypothesis: ['', '▁But', '▁what', "'", 's', '▁interesting', '▁is', '▁the', '▁unique', '▁cadence', '▁of', '▁the', '▁song', ',', '▁the', '▁rhythm', '▁of', '▁the', '▁dance', '▁in', '▁every', '▁culture', '.', '', '▁And', '▁whether', '▁it', "'", 's', '▁the', '▁Pen', 'an', '▁in', '▁the', '▁forests', '▁of', '▁Bo', 'r', 'ne', 'o', ',', '▁or', '▁the', '▁V', 'oodoo', '▁A', 'co', 'ly', 'tes', '▁in', '▁Haiti', ',', '▁or', '▁the', '▁warrior', 's', '▁in', '▁the', '▁northern', '▁Kenya', 'n', '▁desert', ',', '▁the', '▁C', 'ur', 'ende', 'ro', '▁in', '▁the', '▁mountains', '▁of', '▁the', '▁And', 'es', ',', '▁or', '▁a', '▁car', 'ava', 'n', 's', 'he', '▁in', '▁the', '▁middle', '▁of', '▁the', '▁Sahara', '.', '▁And', '▁by', '▁the', '▁way', ',', '▁this', '▁is', '▁the', '▁person', '▁I', '▁traveled', '▁to', '▁the', '▁desert', '▁a', '▁month', '▁ago', ',', '▁or', '▁indeed', ',', '▁the', '▁Yak', "'", 's', '▁her', 'd', 's', '▁on', '▁the', '▁', '▁fla', 'n', 'c', 's', '▁of', '▁Q', 'o', 'mo', 'lang', 'ma', ',', '▁M', 't', '.', '▁Everest', ',', '▁the', '▁global', '▁goddess', '.'] 2024-01-16 16:08:47,392 - INFO - joeynmt.training - Example #2 2024-01-16 16:08:47,393 - INFO - joeynmt.training - Source: Now, together the myriad cultures of the world make up a web of spiritual life and cultural life that envelops the planet, and is as important to the well-being of the planet as indeed is the biological web of life that you know as a biosphere. 2024-01-16 16:08:47,393 - INFO - joeynmt.training - Reference: Aujourd'hui, les innombrables cultures dans le monde constituent un tissu de vie spirituelle et culturelle qui enveloppe la planète, et qui est aussi important pour le bien-être de la planète que l'est également le tissu biologique de la vie que vous connaissez en tant que biosphère. 2024-01-16 16:08:47,393 - INFO - joeynmt.training - Hypothesis: Maintenant, ensemble, les myriades de cultures du monde font un web de vie spirituelle et de vie culturelle qui enveloppent la planète, et est aussi important pour le bien-être de la planète que pour la vérité, le web biologique de la vie que vous connaissez en tant que biosphère. 2024-01-16 16:08:47,394 - DEBUG - joeynmt.training - Tokenized source: ['▁Now', ',', '▁together', '▁the', '▁myriad', '▁cultures', '▁of', '▁the', '▁world', '▁make', '▁up', '▁a', '▁web', '▁of', '▁spiritual', '▁life', '▁and', '▁cultural', '▁life', '▁that', '▁en', 've', 'lop', 's', '▁the', '▁planet', ',', '▁and', '▁is', '▁as', '▁important', '▁to', '▁the', '▁well', '-', 'being', '▁of', '▁the', '▁planet', '▁as', '▁indeed', '▁is', '▁the', '▁biological', '▁web', '▁of', '▁life', '▁that', '▁you', '▁know', '▁as', '▁a', '▁biosphere', '.'] 2024-01-16 16:08:47,394 - DEBUG - joeynmt.training - Tokenized reference: ['▁A', 'ujourd', "'", 'hui', ',', '▁les', '▁', 'innombrables', '▁cultures', '▁dans', '▁le', '▁monde', '▁constituent', '▁un', '▁tissu', '▁de', '▁vie', '▁spirituelle', '▁et', '▁culturelle', '▁qui', '▁enveloppe', '▁la', '▁planète', ',', '▁et', '▁qui', '▁est', '▁aussi', '▁important', '▁pour', '▁le', '▁bien', '-', 'être', '▁de', '▁la', '▁planète', '▁que', '▁l', "'", 'est', '▁également', '▁le', '▁tissu', '▁biologique', '▁de', '▁la', '▁vie', '▁que', '▁vous', '▁connaissez', '▁en', '▁tant', '▁que', '▁biosphère', '.'] 2024-01-16 16:08:47,394 - DEBUG - joeynmt.training - Tokenized hypothesis: ['', '▁Et', '▁ceci', '▁est', '▁une', '▁idée', ',', '▁si', '▁on', '▁y', '▁réfléchi', 't', ',', '▁qui', '▁ne', '▁peut', '▁que', '▁vous', '▁remplir', '▁d', "'", 'espoir', '.', '', '▁Maintenant', ',', '▁ensemble', ',', '▁les', '▁myriad', 'es', '▁de', '▁cultures', '▁du', '▁monde', '▁font', '▁un', '▁web', '▁de', '▁vie', '▁spirituelle', '▁et', '▁de', '▁vie', '▁culturelle', '▁qui', '▁en', 've', 'lop', 'p', 'ent', '▁la', '▁planète', ',', '▁et', '▁est', '▁aussi', '▁important', '▁pour', '▁le', '▁bien', '-', 'être', '▁de', '▁la', '▁planète', '▁que', '▁pour', '▁la', '▁vérité', ',', '▁le', '▁web', '▁biologique', '▁de', '▁la', '▁vie', '▁que', '▁vous', '▁connaissez', '▁en', '▁tant', '▁que', '▁biosphère', '.'] 2024-01-16 16:08:47,394 - INFO - joeynmt.training - Example #3 2024-01-16 16:08:47,395 - INFO - joeynmt.training - Source: And the great indicator of that, of course, is language loss. 2024-01-16 16:08:47,395 - INFO - joeynmt.training - Reference: Et l'indicateur le plus fiable est bien sûr l'extinction du langage. 2024-01-16 16:08:47,395 - INFO - joeynmt.training - Hypothesis: Et le grand indicateur de cela, bien sûr, est la perte de langues. 2024-01-16 16:08:47,396 - DEBUG - joeynmt.training - Tokenized source: ['▁And', '▁the', '▁great', '▁indicator', '▁of', '▁that', ',', '▁of', '▁course', ',', '▁is', '▁language', '▁loss', '.'] 2024-01-16 16:08:47,396 - DEBUG - joeynmt.training - Tokenized reference: ['▁Et', '▁l', "'", 'ind', 'ic', 'ateur', '▁le', '▁plus', '▁fiable', '▁est', '▁bien', '▁sûr', '▁l', "'", 'extinction', '▁du', '▁langage', '.'] 2024-01-16 16:08:47,396 - DEBUG - joeynmt.training - Tokenized hypothesis: ['', '▁Aucun', '▁biologist', 'e', ',', '▁par', '▁exemple', ',', '▁n', "'", 'ose', 'rait', '▁suggérer', '▁que', '▁50%', '▁ou', '▁plus', '▁de', '▁toutes', '▁les', '▁espèces', '▁ont', '▁été', '▁ou', '▁sont', '▁à', '▁deux', '▁doigts', '▁de', '▁l', "'", 'extinction', '▁parce', '▁que', '▁ce', '▁n', "'", 'est', '▁tout', '▁simplement', '▁pas', '▁vrai', ',', '▁et', '▁pourtant', '▁--', '▁que', '▁le', '▁scénario', '▁le', '▁plus', '▁a', 'po', 'ca', 'ly', 'p', 'tique', '▁dans', '▁le', '▁royaume', '▁de', '▁la', '▁diversité', '▁biologique', '▁--', '▁se', '▁rapproche', '▁rarement', '▁de', '▁ce', '▁que', '▁nous', '▁considér', 'ons', '▁comme', '▁le', '▁scénario', '▁le', '▁plus', '▁optimiste', '▁au', '▁sein', '▁de', '▁la', '▁diversité', '▁culturelle', '.', '', '▁Et', '▁le', '▁grand', '▁indicate', 'ur', '▁de', '▁cela', ',', '▁bien', '▁sûr', ',', '▁est', '▁la', '▁perte', '▁de', '▁langues', '.'] 2024-01-16 16:10:03,139 - INFO - joeynmt.training - Epoch 11, Step: 28100, Batch Loss: 0.881707, Batch Acc: 0.767043, Tokens per Sec: 16433, Lr: 0.000119 2024-01-16 16:11:19,235 - INFO - joeynmt.training - Epoch 11, Step: 28200, Batch Loss: 0.903995, Batch Acc: 0.765882, Tokens per Sec: 16409, Lr: 0.000119 2024-01-16 16:12:36,209 - INFO - joeynmt.training - Epoch 11, Step: 28300, Batch Loss: 0.880206, Batch Acc: 0.766191, Tokens per Sec: 16281, Lr: 0.000119 2024-01-16 16:13:51,929 - INFO - joeynmt.training - Epoch 11, Step: 28400, Batch Loss: 0.938306, Batch Acc: 0.764872, Tokens per Sec: 16440, Lr: 0.000119 2024-01-16 16:15:07,387 - INFO - joeynmt.training - Epoch 11, Step: 28500, Batch Loss: 0.905782, Batch Acc: 0.766254, Tokens per Sec: 16468, Lr: 0.000118 2024-01-16 16:16:22,804 - INFO - joeynmt.training - Epoch 11, Step: 28600, Batch Loss: 0.892838, Batch Acc: 0.765772, Tokens per Sec: 16512, Lr: 0.000118 2024-01-16 16:17:38,508 - INFO - joeynmt.training - Epoch 11, Step: 28700, Batch Loss: 0.942997, Batch Acc: 0.766366, Tokens per Sec: 16443, Lr: 0.000118 2024-01-16 16:18:54,130 - INFO - joeynmt.training - Epoch 11, Step: 28800, Batch Loss: 0.914250, Batch Acc: 0.763783, Tokens per Sec: 16555, Lr: 0.000118 2024-01-16 16:20:10,121 - INFO - joeynmt.training - Epoch 11, Step: 28900, Batch Loss: 0.924553, Batch Acc: 0.765473, Tokens per Sec: 16446, Lr: 0.000118 2024-01-16 16:21:26,717 - INFO - joeynmt.training - Epoch 11, Step: 29000, Batch Loss: 0.885420, Batch Acc: 0.766206, Tokens per Sec: 16359, Lr: 0.000117 2024-01-16 16:21:26,730 - INFO - joeynmt.datasets - Sample random subset from dev data: n=500, seed=29042 2024-01-16 16:21:26,730 - INFO - joeynmt.prediction - Predicting 500 example(s)... (Greedy decoding with min_output_length=1, max_output_length=512, return_prob='none', generate_unk=True, repetition_penalty=-1, no_repeat_ngram_size=-1) 2024-01-16 16:21:58,945 - INFO - joeynmt.prediction - Generation took 32.2139[sec]. 2024-01-16 16:21:59,150 - INFO - joeynmt.metrics - nrefs:1|case:lc|eff:no|tok:13a|smooth:exp|version:2.4.0 2024-01-16 16:21:59,151 - INFO - joeynmt.prediction - Evaluation result (greedy): bleu: 25.79, loss: 2.10, ppl: 8.15, acc: 0.59, 0.0707[sec] 2024-01-16 16:21:59,152 - INFO - joeynmt.training - Example #0 2024-01-16 16:21:59,152 - INFO - joeynmt.training - Source: You know, one of the intense pleasures of travel and one of the delights of ethnographic research is the opportunity to live amongst those who have not forgotten the old ways, who still feel their past in the wind, touch it in stones polished by rain, taste it in the bitter leaves of plants. 2024-01-16 16:21:59,153 - INFO - joeynmt.training - Reference: Vous savez, un des plaisirs intenses du voyage et un des délices de la recherche ethnographique est la possibilité de vivre parmi ceux qui n'ont pas oublié les anciennes coutumes, qui ressentent encore leur passé souffler dans le vent, qui le touchent dans les pierres polies par la pluie, le dégustent dans les feuilles amères des plantes. 2024-01-16 16:21:59,153 - INFO - joeynmt.training - Hypothesis: Vous savez, l'un des plaisirs intenses du voyage et l'un des plaisirs de recherche ethnographique est l'occasion de vivre parmi ceux qui n'ont pas oublié les anciennes façons, qui ont encore le temps dans le vent, le toucher dans des pierres polies par la pluie, le goûter dans les laves amerantes des plantes. 2024-01-16 16:21:59,154 - DEBUG - joeynmt.training - Tokenized source: ['▁You', '▁know', ',', '▁one', '▁of', '▁the', '▁intense', '▁pleasure', 's', '▁of', '▁travel', '▁and', '▁one', '▁of', '▁the', '▁de', 'light', 's', '▁of', '▁et', 'h', 'n', 'ographic', '▁research', '▁is', '▁the', '▁opportunity', '▁to', '▁live', '▁among', 'st', '▁those', '▁who', '▁have', '▁not', '▁forgotten', '▁the', '▁old', '▁ways', ',', '▁who', '▁still', '▁feel', '▁their', '▁past', '▁in', '▁the', '▁wind', ',', '▁touch', '▁it', '▁in', '▁stones', '▁polish', 'ed', '▁by', '▁rain', ',', '▁taste', '▁it', '▁in', '▁the', '▁bitter', '▁leaves', '▁of', '▁plants', '.'] 2024-01-16 16:21:59,154 - DEBUG - joeynmt.training - Tokenized reference: ['▁Vous', '▁savez', ',', '▁un', '▁des', '▁plaisir', 's', '▁intense', 's', '▁du', '▁voyage', '▁et', '▁un', '▁des', '▁dé', 'lic', 'es', '▁de', '▁la', '▁recherche', '▁et', 'h', 'no', 'graph', 'ique', '▁est', '▁la', '▁possibilité', '▁de', '▁vivre', '▁parmi', '▁ceux', '▁qui', '▁n', "'", 'ont', '▁pas', '▁oublié', '▁les', '▁anciennes', '▁coutume', 's', ',', '▁qui', '▁ressentent', '▁encore', '▁leur', '▁passé', '▁souffle', 'r', '▁dans', '▁le', '▁vent', ',', '▁qui', '▁le', '▁touchent', '▁dans', '▁les', '▁pierres', '▁poli', 'es', '▁par', '▁la', '▁pluie', ',', '▁le', '▁dé', 'gu', 'sten', 't', '▁dans', '▁les', '▁feuilles', '▁a', 'mère', 's', '▁des', '▁plantes', '.'] 2024-01-16 16:21:59,154 - DEBUG - joeynmt.training - Tokenized hypothesis: ['', '', '▁Vous', '▁savez', ',', '▁l', "'", 'un', '▁des', '▁plaisir', 's', '▁intense', 's', '▁du', '▁voyage', '▁et', '▁l', "'", 'un', '▁des', '▁plaisir', 's', '▁de', '▁recherche', '▁et', 'h', 'no', 'graph', 'ique', '▁est', '▁l', "'", 'occasion', '▁de', '▁vivre', '▁parmi', '▁ceux', '▁qui', '▁n', "'", 'ont', '▁pas', '▁oublié', '▁les', '▁anciennes', '▁façons', ',', '▁qui', '▁ont', '▁encore', '▁le', '▁temps', '▁dans', '▁le', '▁vent', ',', '▁le', '▁toucher', '▁dans', '▁des', '▁pierres', '▁poli', 'es', '▁par', '▁la', '▁pluie', ',', '▁le', '▁goût', 'er', '▁dans', '▁les', '▁la', 've', 's', '▁am', 'er', 'antes', '▁des', '▁plantes', '.'] 2024-01-16 16:21:59,154 - INFO - joeynmt.training - Example #1 2024-01-16 16:21:59,154 - INFO - joeynmt.training - Source: Et vous pourriez considérer ce tissu culturel de la vie en tant qu'ethnosphère et vous pourriez définir l'ethnosphère comme étant la somme globale de toutes les pensées, les rêves, les mythes, les idées, les inspirations, les intuitions engendrées par l'imagination humaine depuis l'aube de la conscience. 2024-01-16 16:21:59,155 - INFO - joeynmt.training - Reference: And you might think of this cultural web of life as being an ethnosphere, and you might define the ethnosphere as being the sum total of all thoughts and dreams, myths, ideas, inspirations, intuitions brought into being by the human imagination since the dawn of consciousness. 2024-01-16 16:21:59,155 - INFO - joeynmt.training - Hypothesis: And you might think of this cultural fabric of life as an agnostic, and you might think of the agnostic as the overall sum of all thoughts, dreams, myths, ideas, inspirations, intuitions that come from the human imagination from the dawn of consciousness. 2024-01-16 16:21:59,156 - DEBUG - joeynmt.training - Tokenized source: ['▁Et', '▁vous', '▁pourriez', '▁considérer', '▁ce', '▁tissu', '▁culturel', '▁de', '▁la', '▁vie', '▁en', '▁tant', '▁qu', "'", 'e', 'th', 'no', 's', 'ph', 'ère', '▁et', '▁vous', '▁pourriez', '▁définir', '▁l', "'", 'e', 'th', 'no', 's', 'ph', 'ère', '▁comme', '▁étant', '▁la', '▁somme', '▁globale', '▁de', '▁toutes', '▁les', '▁pensées', ',', '▁les', '▁rêves', ',', '▁les', '▁mythe', 's', ',', '▁les', '▁idées', ',', '▁les', '▁inspiration', 's', ',', '▁les', '▁intuition', 's', '▁engendré', 'es', '▁par', '▁l', "'", 'imagination', '▁humaine', '▁depuis', '▁l', "'", 'aube', '▁de', '▁la', '▁conscience', '.'] 2024-01-16 16:21:59,156 - DEBUG - joeynmt.training - Tokenized reference: ['▁And', '▁you', '▁might', '▁think', '▁of', '▁this', '▁cultural', '▁web', '▁of', '▁life', '▁as', '▁being', '▁an', '▁et', 'h', 'n', 'osphere', ',', '▁and', '▁you', '▁might', '▁define', '▁the', '▁et', 'h', 'n', 'osphere', '▁as', '▁being', '▁the', '▁sum', '▁total', '▁of', '▁all', '▁thoughts', '▁and', '▁dreams', ',', '▁myth', 's', ',', '▁ideas', ',', '▁inspiration', 's', ',', '▁intuition', 's', '▁brought', '▁into', '▁being', '▁by', '▁the', '▁human', '▁imagination', '▁since', '▁the', '▁dawn', '▁of', '▁consciousness', '.'] 2024-01-16 16:21:59,156 - DEBUG - joeynmt.training - Tokenized hypothesis: ['', '▁Now', ',', '▁together', '▁the', '▁myriad', '▁cultures', '▁of', '▁the', '▁world', '▁make', '▁up', '▁a', '▁web', '▁of', '▁spiritual', '▁life', '▁and', '▁cultural', '▁life', '▁that', '▁en', 've', 'lop', 's', '▁the', '▁planet', ',', '▁and', '▁is', '▁as', '▁important', '▁to', '▁the', '▁well', '-', 'being', '▁of', '▁the', '▁planet', '▁as', '▁indeed', '▁is', '▁the', '▁biological', '▁web', '▁of', '▁life', '▁that', '▁you', '▁know', '▁as', '▁a', '▁biosphere', '.', '', '▁And', '▁you', '▁might', '▁think', '▁of', '▁this', '▁cultural', '▁fabric', '▁of', '▁life', '▁as', '▁an', '▁a', 'gnostic', ',', '▁and', '▁you', '▁might', '▁think', '▁of', '▁the', '▁a', 'gnostic', '▁as', '▁the', '▁overall', '▁sum', '▁of', '▁all', '▁thoughts', ',', '▁dreams', ',', '▁myth', 's', ',', '▁ideas', ',', '▁inspiration', 's', ',', '▁intuition', 's', '▁that', '▁come', '▁from', '▁the', '▁human', '▁imagination', '▁from', '▁the', '▁dawn', '▁of', '▁consciousness', '.'] 2024-01-16 16:21:59,156 - INFO - joeynmt.training - Example #2 2024-01-16 16:21:59,156 - INFO - joeynmt.training - Source: The ethnosphere is humanity's great legacy. 2024-01-16 16:21:59,156 - INFO - joeynmt.training - Reference: L'ethnosphère est l'héritage de l'humanité. 2024-01-16 16:21:59,156 - INFO - joeynmt.training - Hypothesis: L'éthémosphäre est l'héritage magnifique de l'humanité. 2024-01-16 16:21:59,157 - DEBUG - joeynmt.training - Tokenized source: ['▁The', '▁et', 'h', 'n', 'osphere', '▁is', '▁humanity', "'", 's', '▁great', '▁legacy', '.'] 2024-01-16 16:21:59,157 - DEBUG - joeynmt.training - Tokenized reference: ['▁L', "'", 'e', 'th', 'no', 's', 'ph', 'ère', '▁est', '▁l', "'", 'héritage', '▁de', '▁l', "'", 'humanité', '.'] 2024-01-16 16:21:59,157 - DEBUG - joeynmt.training - Tokenized hypothesis: ['', '▁Et', '▁vous', '▁pourriez', '▁considérer', '▁ce', '▁tissu', '▁culturel', '▁de', '▁la', '▁vie', '▁en', '▁tant', '▁qu', "'", 'e', 'th', 'no', 's', 'ph', 'ère', '▁et', '▁vous', '▁pourriez', '▁définir', '▁l', "'", 'e', 'th', 'no', 's', 'ph', 'ère', '▁comme', '▁étant', '▁la', '▁somme', '▁globale', '▁de', '▁toutes', '▁les', '▁pensées', ',', '▁les', '▁rêves', ',', '▁les', '▁mythe', 's', ',', '▁les', '▁idées', ',', '▁les', '▁inspiration', 's', ',', '▁les', '▁intuition', 's', '▁engendré', 'es', '▁par', '▁l', "'", 'imagination', '▁humaine', '▁depuis', '▁l', "'", 'aube', '▁de', '▁la', '▁conscience', '.', '', '▁L', "'", 'é', 't', 'hé', 'mo', 'sphäre', '▁est', '▁l', "'", 'héritage', '▁magnifique', '▁de', '▁l', "'", 'humanité', '.'] 2024-01-16 16:21:59,157 - INFO - joeynmt.training - Example #3 2024-01-16 16:21:59,158 - INFO - joeynmt.training - Source: Et je sais que certains d'entre vous disent, "Ne serait-il pas mieux ? Le monde ne serait-il pas un meilleur endroit si nous ne parlions qu'une seule langue ?" Et je réponds, "Bien, cette langue sera du Yoruba. Du Cantonais. 2024-01-16 16:21:59,158 - INFO - joeynmt.training - Reference: And I know there's some of you who say, "Well, wouldn't it be better, wouldn't the world be a better place if we all just spoke one language?" And I say, "Great, let's make that language Yoruba. Let's make it Cantonese. 2024-01-16 16:21:59,158 - INFO - joeynmt.training - Hypothesis: And I know some of you are saying, "Wouldn't it be better? Wouldn't the world be better if we were just talking one language?" And I said, "Well, that language would be Yoruba. Cantones. 2024-01-16 16:21:59,159 - DEBUG - joeynmt.training - Tokenized source: ['▁Et', '▁je', '▁sais', '▁que', '▁certains', '▁d', "'", 'entre', '▁vous', '▁disent', ',', '▁"', 'Ne', '▁serait', '-', 'il', '▁pas', '▁mieux', '▁?', '▁Le', '▁monde', '▁ne', '▁serait', '-', 'il', '▁pas', '▁un', '▁meilleur', '▁endroit', '▁si', '▁nous', '▁ne', '▁parlions', '▁qu', "'", 'une', '▁seule', '▁langue', '▁?"', '▁Et', '▁je', '▁répond', 's', ',', '▁"', 'Bien', ',', '▁cette', '▁langue', '▁sera', '▁du', '▁Yoruba', '.', '▁Du', '▁Can', 'ton', 'ais', '.'] 2024-01-16 16:21:59,159 - DEBUG - joeynmt.training - Tokenized reference: ['▁And', '▁I', '▁know', '▁there', "'", 's', '▁some', '▁of', '▁you', '▁who', '▁say', ',', '▁"', 'Well', ',', '▁would', 'n', "'", 't', '▁it', '▁be', '▁better', ',', '▁would', 'n', "'", 't', '▁the', '▁world', '▁be', '▁a', '▁better', '▁place', '▁if', '▁we', '▁all', '▁just', '▁spoke', '▁one', '▁language', '?"', '▁And', '▁I', '▁say', ',', '▁"', 'Great', ',', '▁let', "'", 's', '▁make', '▁that', '▁language', '▁Yoruba', '.', '▁Let', "'", 's', '▁make', '▁it', '▁Can', 'ton', 'es', 'e', '.'] 2024-01-16 16:21:59,159 - DEBUG - joeynmt.training - Tokenized hypothesis: ['', '▁And', '▁yet', ',', '▁that', '▁dreadful', '▁fate', '▁is', '▁indeed', '▁the', '▁p', 'light', '▁of', '▁somebody', '▁somewhere', '▁on', '▁Earth', '▁roughly', '▁every', '▁two', '▁weeks', ',', '▁because', '▁every', '▁two', '▁weeks', ',', '▁some', '▁el', 'der', '▁die', 's', '▁and', '▁carries', '▁with', '▁him', '▁into', '▁the', '▁grave', '▁the', '▁last', '▁s', 'y', 'll', 'ables', '▁of', '▁an', '▁ancient', '▁tongue', '.', '', '▁And', '▁I', '▁know', '▁some', '▁of', '▁you', '▁are', '▁saying', ',', '▁"', 'Would', 'n', "'", 't', '▁it', '▁be', '▁better', '?', '▁Would', 'n', "'", 't', '▁the', '▁world', '▁be', '▁better', '▁if', '▁we', '▁were', '▁just', '▁talking', '▁one', '▁language', '?"', '▁And', '▁I', '▁said', ',', '▁"', 'Well', ',', '▁that', '▁language', '▁would', '▁be', '▁Yoruba', '.', '▁Can', 'ton', 'es', '.'] 2024-01-16 16:23:14,314 - INFO - joeynmt.training - Epoch 11, Step: 29100, Batch Loss: 0.950500, Batch Acc: 0.765358, Tokens per Sec: 16611, Lr: 0.000117 2024-01-16 16:24:30,322 - INFO - joeynmt.training - Epoch 11, Step: 29200, Batch Loss: 0.924802, Batch Acc: 0.765183, Tokens per Sec: 16440, Lr: 0.000117 2024-01-16 16:25:18,743 - INFO - joeynmt.training - Interrupt at epoch 11, step 29264. 2024-01-16 16:25:21,529 - INFO - joeynmt.training - Checkpoint saved in models/iwslt14_prompt/29264.ckpt. 2024-01-16 16:25:21,532 - INFO - joeynmt.training - Skipping test after training.