End of training

README.md

@@ -18,11 +18,11 @@ should probably proofread and complete it, then remove this comment. -->
 
 This model is a fine-tuned version of [aubmindlab/aragpt2-large](https://huggingface.co/aubmindlab/aragpt2-large) on an unknown dataset.
 It achieves the following results on the evaluation set:
-- Loss: 0.0930
-- Bleu: 0.1326
-- Rouge1: 0.4960
-- Rouge2: 0.2713
-- Rougel: 0.4923
+- Loss: 0.0718
+- Bleu: 0.1063
+- Rouge1: 0.4552
+- Rouge2: 0.2232
+- Rougel: 0.4516
 
 ## Model description
 

all_results.json

@@ -0,0 +1,19 @@
+{
+    "epoch": 7.0,
+    "eval_bleu": 0.10627535338849203,
+    "eval_loss": 0.07183855026960373,
+    "eval_rouge1": 0.45517089056426413,
+    "eval_rouge2": 0.2232454846856679,
+    "eval_rougeL": 0.4515800058148628,
+    "eval_runtime": 828.4394,
+    "eval_samples": 5380,
+    "eval_samples_per_second": 6.494,
+    "eval_steps_per_second": 1.624,
+    "perplexity": 1.0744818552534834,
+    "total_flos": 6.528656496918528e+17,
+    "train_loss": 0.0533416826159377,
+    "train_runtime": 67570.4522,
+    "train_samples": 21429,
+    "train_samples_per_second": 6.343,
+    "train_steps_per_second": 1.586
+}

egy_training_log.txt

@@ -568,3 +568,5 @@ INFO:root:Epoch 6.0: Train Loss = 0.0325, Eval Loss = 0.08186400681734085
 INFO:absl:Using default tokenizer.
 INFO:root:Epoch 7.0: Train Loss = 0.0262, Eval Loss = 0.08628461509943008
 INFO:absl:Using default tokenizer.
+INFO:__main__:*** Evaluate ***
+INFO:absl:Using default tokenizer.

eval_results.json

@@ -0,0 +1,13 @@
+{
+    "epoch": 7.0,
+    "eval_bleu": 0.10627535338849203,
+    "eval_loss": 0.07183855026960373,
+    "eval_rouge1": 0.45517089056426413,
+    "eval_rouge2": 0.2232454846856679,
+    "eval_rougeL": 0.4515800058148628,
+    "eval_runtime": 828.4394,
+    "eval_samples": 5380,
+    "eval_samples_per_second": 6.494,
+    "eval_steps_per_second": 1.624,
+    "perplexity": 1.0744818552534834
+}

train_results.json

@@ -0,0 +1,9 @@
+{
+    "epoch": 7.0,
+    "total_flos": 6.528656496918528e+17,
+    "train_loss": 0.0533416826159377,
+    "train_runtime": 67570.4522,
+    "train_samples": 21429,
+    "train_samples_per_second": 6.343,
+    "train_steps_per_second": 1.586
+}

trainer_state.json

@@ -0,0 +1,184 @@
+{
+  "best_metric": 0.07183855026960373,
+  "best_model_checkpoint": "/home/iais_marenpielka/Bouthaina/res_nw_dj_aragpt2-large/checkpoint-10716",
+  "epoch": 7.0,
+  "eval_steps": 500,
+  "global_step": 37506,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 1.0,
+      "grad_norm": 1.1402301788330078,
+      "learning_rate": 4.772267016688543e-05,
+      "loss": 0.1323,
+      "step": 5358
+    },
+    {
+      "epoch": 1.0,
+      "eval_bleu": 0.08300663646266425,
+      "eval_loss": 0.07568918913602829,
+      "eval_rouge1": 0.4072217751588082,
+      "eval_rouge2": 0.18254715184887235,
+      "eval_rougeL": 0.4034974548012786,
+      "eval_runtime": 706.7922,
+      "eval_samples_per_second": 7.612,
+      "eval_steps_per_second": 1.903,
+      "step": 5358
+    },
+    {
+      "epoch": 2.0,
+      "grad_norm": 0.7255021929740906,
+      "learning_rate": 4.521095068441778e-05,
+      "loss": 0.0662,
+      "step": 10716
+    },
+    {
+      "epoch": 2.0,
+      "eval_bleu": 0.10627535338849203,
+      "eval_loss": 0.07183855026960373,
+      "eval_rouge1": 0.45517089056426413,
+      "eval_rouge2": 0.2232454846856679,
+      "eval_rougeL": 0.4515800058148628,
+      "eval_runtime": 715.8215,
+      "eval_samples_per_second": 7.516,
+      "eval_steps_per_second": 1.879,
+      "step": 10716
+    },
+    {
+      "epoch": 3.0,
+      "grad_norm": 0.5735991597175598,
+      "learning_rate": 4.269923120195012e-05,
+      "loss": 0.0526,
+      "step": 16074
+    },
+    {
+      "epoch": 3.0,
+      "eval_bleu": 0.11970864457422921,
+      "eval_loss": 0.07271187007427216,
+      "eval_rouge1": 0.47525716850104355,
+      "eval_rouge2": 0.25197591385088847,
+      "eval_rougeL": 0.47186762638658064,
+      "eval_runtime": 766.9396,
+      "eval_samples_per_second": 7.015,
+      "eval_steps_per_second": 1.754,
+      "step": 16074
+    },
+    {
+      "epoch": 4.0,
+      "grad_norm": 0.9193664789199829,
+      "learning_rate": 4.018751171948247e-05,
+      "loss": 0.0414,
+      "step": 21432
+    },
+    {
+      "epoch": 4.0,
+      "eval_bleu": 0.12737868761714774,
+      "eval_loss": 0.07567169517278671,
+      "eval_rouge1": 0.4894464779386084,
+      "eval_rouge2": 0.2644220876804199,
+      "eval_rougeL": 0.48620236061532013,
+      "eval_runtime": 767.056,
+      "eval_samples_per_second": 7.014,
+      "eval_steps_per_second": 1.753,
+      "step": 21432
+    },
+    {
+      "epoch": 5.0,
+      "grad_norm": 1.2567533254623413,
+      "learning_rate": 3.7675792237014815e-05,
+      "loss": 0.0325,
+      "step": 26790
+    },
+    {
+      "epoch": 5.0,
+      "eval_bleu": 0.12904171218936847,
+      "eval_loss": 0.08186400681734085,
+      "eval_rouge1": 0.4910163857094879,
+      "eval_rouge2": 0.26712580782902845,
+      "eval_rougeL": 0.4874926457795191,
+      "eval_runtime": 767.2369,
+      "eval_samples_per_second": 7.012,
+      "eval_steps_per_second": 1.753,
+      "step": 26790
+    },
+    {
+      "epoch": 6.0,
+      "grad_norm": 1.125952124595642,
+      "learning_rate": 3.516407275454716e-05,
+      "loss": 0.0262,
+      "step": 32148
+    },
+    {
+      "epoch": 6.0,
+      "eval_bleu": 0.12974770289755774,
+      "eval_loss": 0.08628461509943008,
+      "eval_rouge1": 0.4922385680285879,
+      "eval_rouge2": 0.26649486337370165,
+      "eval_rougeL": 0.4887932180100094,
+      "eval_runtime": 828.4635,
+      "eval_samples_per_second": 6.494,
+      "eval_steps_per_second": 1.623,
+      "step": 32148
+    },
+    {
+      "epoch": 7.0,
+      "grad_norm": 0.9687314629554749,
+      "learning_rate": 3.26523532720795e-05,
+      "loss": 0.0221,
+      "step": 37506
+    },
+    {
+      "epoch": 7.0,
+      "eval_bleu": 0.13258665968806682,
+      "eval_loss": 0.09302929788827896,
+      "eval_rouge1": 0.4960344976284088,
+      "eval_rouge2": 0.2712943985267227,
+      "eval_rougeL": 0.4923109277459174,
+      "eval_runtime": 828.274,
+      "eval_samples_per_second": 6.495,
+      "eval_steps_per_second": 1.624,
+      "step": 37506
+    },
+    {
+      "epoch": 7.0,
+      "step": 37506,
+      "total_flos": 6.528656496918528e+17,
+      "train_loss": 0.0533416826159377,
+      "train_runtime": 67570.4522,
+      "train_samples_per_second": 6.343,
+      "train_steps_per_second": 1.586
+    }
+  ],
+  "logging_steps": 500,
+  "max_steps": 107160,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 20,
+  "save_steps": 500,
+  "stateful_callbacks": {
+    "EarlyStoppingCallback": {
+      "args": {
+        "early_stopping_patience": 5,
+        "early_stopping_threshold": 0.0
+      },
+      "attributes": {
+        "early_stopping_patience_counter": 0
+      }
+    },
+    "TrainerControl": {
+      "args": {
+        "should_epoch_stop": false,
+        "should_evaluate": false,
+        "should_log": false,
+        "should_save": true,
+        "should_training_stop": true
+      },
+      "attributes": {}
+    }
+  },
+  "total_flos": 6.528656496918528e+17,
+  "train_batch_size": 4,
+  "trial_name": null,
+  "trial_params": null
+}