End of training

README.md

@@ -1,6 +1,6 @@
 ---
 license: apache-2.0
-base_model: harshit345/xlsr-wav2vec-speech-emotion-recognition
+base_model: facebook/hubert-base-ls960
 tags:
 - generated_from_trainer
 metrics:
@@ -15,10 +15,10 @@ should probably proofread and complete it, then remove this comment. -->
 
 # ckpts
 
-This model is a fine-tuned version of [harshit345/xlsr-wav2vec-speech-emotion-recognition](https://huggingface.co/harshit345/xlsr-wav2vec-speech-emotion-recognition) on an unknown dataset.
+This model is a fine-tuned version of [facebook/hubert-base-ls960](https://huggingface.co/facebook/hubert-base-ls960) on an unknown dataset.
 It achieves the following results on the evaluation set:
-- Loss: 0.2342
-- Accuracy: 0.9333
+- Loss: 0.2499
+- Accuracy: 0.9657
 
 ## Model description
 
@@ -45,15 +45,17 @@ The following hyperparameters were used during training:
 - total_train_batch_size: 8
 - optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
 - lr_scheduler_type: linear
-- num_epochs: 5.0
+- num_epochs: 10.0
 - mixed_precision_training: Native AMP
 
 ### Training results
 
 | Training Loss | Epoch | Step | Validation Loss | Accuracy |
 |:-------------:|:-----:|:----:|:---------------:|:--------:|
-| 0.4039        | 2.02  | 500  | 0.3198          | 0.8949   |
-| 0.2195        | 4.04  | 1000 | 0.2363          | 0.9333   |
+| 0.1792        | 2.02  | 500  | 0.2998          | 0.9313   |
+| 0.1097        | 4.04  | 1000 | 0.3281          | 0.9475   |
+| 0.0557        | 6.06  | 1500 | 0.2455          | 0.9657   |
+| 0.0535        | 8.08  | 2000 | 0.2268          | 0.9657   |
 
 
 ### Framework versions

all_results.json

@@ -1,9 +1,9 @@
 {
-    "epoch": 4.99,
-    "total_flos": 1.053348926143392e+18,
-    "train_loss": 0.5674189640925481,
-    "train_runtime": 1664.3033,
+    "epoch": 9.98,
+    "total_flos": 6.32707029472752e+17,
+    "train_loss": 0.09667120742411749,
+    "train_runtime": 1319.7559,
     "train_samples": 1979,
-    "train_samples_per_second": 5.945,
-    "train_steps_per_second": 0.742
+    "train_samples_per_second": 14.995,
+    "train_steps_per_second": 1.872
 }

config.json

@@ -1,20 +1,14 @@
 {
-  "_name_or_path": "harshit345/xlsr-wav2vec-speech-emotion-recognition",
-  "activation_dropout": 0.0,
-  "adapter_attn_dim": null,
-  "adapter_kernel_size": 3,
-  "adapter_stride": 2,
-  "add_adapter": false,
+  "_name_or_path": "facebook/hubert-base-ls960",
+  "activation_dropout": 0.1,
   "apply_spec_augment": true,
   "architectures": [
-    "Wav2Vec2ForSpeechClassification"
+    "HubertForSpeechClassification"
   ],
   "attention_dropout": 0.1,
   "bos_token_id": 1,
   "classifier_proj_size": 256,
-  "codevector_dim": 256,
-  "contrastive_logits_temperature": 0.1,
-  "conv_bias": true,
+  "conv_bias": false,
   "conv_dim": [
     512,
     512,
@@ -42,21 +36,22 @@
     2,
     2
   ],
-  "ctc_loss_reduction": "mean",
-  "ctc_zero_infinity": true,
-  "diversity_loss_weight": 0.1,
-  "do_stable_layer_norm": true,
+  "ctc_loss_reduction": "sum",
+  "ctc_zero_infinity": false,
+  "do_stable_layer_norm": false,
   "eos_token_id": 2,
   "feat_extract_activation": "gelu",
   "feat_extract_dropout": 0.0,
-  "feat_extract_norm": "layer",
-  "feat_proj_dropout": 0.0,
-  "feat_quantizer_dropout": 0.0,
-  "final_dropout": 0.0,
+  "feat_extract_norm": "group",
+  "feat_proj_dropout": 0.1,
+  "feat_proj_layer_norm": true,
+  "final_dropout": 0.1,
   "finetuning_task": "wav2vec2_clf",
+  "gradient_checkpointing": false,
   "hidden_act": "gelu",
   "hidden_dropout": 0.1,
-  "hidden_size": 1024,
+  "hidden_dropout_prob": 0.1,
+  "hidden_size": 768,
   "id2label": {
     "0": "01Neutral",
     "1": "02Fearful",
@@ -65,7 +60,7 @@
     "4": "05Angry"
   },
   "initializer_range": 0.02,
-  "intermediate_size": 4096,
+  "intermediate_size": 3072,
   "label2id": {
     "01Neutral": 0,
     "02Fearful": 1,
@@ -75,59 +70,24 @@
   },
   "layer_norm_eps": 1e-05,
   "layerdrop": 0.1,
-  "mask_channel_length": 10,
-  "mask_channel_min_space": 1,
-  "mask_channel_other": 0.0,
-  "mask_channel_prob": 0.0,
-  "mask_channel_selection": "static",
   "mask_feature_length": 10,
   "mask_feature_min_masks": 0,
   "mask_feature_prob": 0.0,
   "mask_time_length": 10,
   "mask_time_min_masks": 2,
-  "mask_time_min_space": 1,
-  "mask_time_other": 0.0,
   "mask_time_prob": 0.05,
-  "mask_time_selection": "static",
-  "model_type": "wav2vec2",
-  "num_adapter_layers": 3,
-  "num_attention_heads": 16,
-  "num_codevector_groups": 2,
-  "num_codevectors_per_group": 320,
+  "model_type": "hubert",
+  "num_attention_heads": 12,
   "num_conv_pos_embedding_groups": 16,
   "num_conv_pos_embeddings": 128,
   "num_feat_extract_layers": 7,
-  "num_hidden_layers": 24,
-  "num_negatives": 100,
-  "output_hidden_size": 1024,
-  "pad_token_id": 54,
+  "num_hidden_layers": 12,
+  "pad_token_id": 0,
   "pooling_mode": "mean",
   "problem_type": "single_label_classification",
-  "proj_codevector_dim": 256,
-  "tdnn_dilation": [
-    1,
-    2,
-    3,
-    1,
-    1
-  ],
-  "tdnn_dim": [
-    512,
-    512,
-    512,
-    512,
-    1500
-  ],
-  "tdnn_kernel": [
-    5,
-    3,
-    3,
-    1,
-    1
-  ],
+  "tokenizer_class": "Wav2Vec2CTCTokenizer",
   "torch_dtype": "float32",
   "transformers_version": "4.40.0.dev0",
   "use_weighted_layer_sum": false,
-  "vocab_size": 55,
-  "xvector_output_dim": 512
+  "vocab_size": 32
 }

model.safetensors

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:4c211db44690f250de2ef2b76fda76600205ce3558af66c9bc6dea154e1a0342
-size 1266026604
+oid sha256:3b2a6e0278ae6a2cdd290bf7567377ae180097bb1ffded26a6f8af803546df0f
+size 379890236

preprocessor_config.json

@@ -3,7 +3,7 @@
   "feature_extractor_type": "Wav2Vec2FeatureExtractor",
   "feature_size": 1,
   "padding_side": "right",
-  "padding_value": 0.0,
-  "return_attention_mask": true,
+  "padding_value": 0,
+  "return_attention_mask": false,
   "sampling_rate": 16000
 }

runs/Apr13_10-12-52_06e814d255dd/events.out.tfevents.1713003179.06e814d255dd.34.0

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:14593a4674391536a6f9675fca55ce223ccfa7fd5e05963dd409f71c246ae69d
+size 5854

runs/Apr13_10-14-18_06e814d255dd/events.out.tfevents.1713003267.06e814d255dd.34.1

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:3bb727ac50279f47365f9d0c30c1b54964736d11c7468437544e611f1c946244
+size 12560

runs/Apr13_10-14-18_06e814d255dd/events.out.tfevents.1713004577.06e814d255dd.34.2

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:09b9d3cdb24b42204ebedf81afc0b8d1dddc37a860fd10acd2771e31275ebf41
+size 12655

runs/Apr13_10-14-18_06e814d255dd/events.out.tfevents.1713005927.06e814d255dd.34.3

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:83ed2b5922a117f368bd2b98fc88d8dec1a8dfc25a01a30021fc26ae702cfc24
+size 411

train_results.json

@@ -1,9 +1,9 @@
 {
-    "epoch": 4.99,
-    "total_flos": 1.053348926143392e+18,
-    "train_loss": 0.5674189640925481,
-    "train_runtime": 1664.3033,
+    "epoch": 9.98,
+    "total_flos": 6.32707029472752e+17,
+    "train_loss": 0.09667120742411749,
+    "train_runtime": 1319.7559,
     "train_samples": 1979,
-    "train_samples_per_second": 5.945,
-    "train_steps_per_second": 0.742
+    "train_samples_per_second": 14.995,
+    "train_steps_per_second": 1.872
 }

trainer_state.json

@@ -1,131 +1,233 @@
 {
   "best_metric": null,
   "best_model_checkpoint": null,
-  "epoch": 4.98989898989899,
+  "epoch": 9.97979797979798,
   "eval_steps": 500,
-  "global_step": 1235,
+  "global_step": 2470,
   "is_hyper_param_search": false,
   "is_local_process_zero": true,
   "is_world_process_zero": true,
   "log_history": [
     {
       "epoch": 0.4,
-      "grad_norm": 4.1816935539245605,
-      "learning_rate": 9.206477732793523e-06,
-      "loss": 2.284,
+      "grad_norm": 55.51200485229492,
+      "learning_rate": 9.595141700404859e-06,
+      "loss": 0.1966,
       "step": 100
     },
     {
       "epoch": 0.81,
-      "grad_norm": 7.234564304351807,
-      "learning_rate": 8.39676113360324e-06,
-      "loss": 1.1378,
+      "grad_norm": 0.044024500995874405,
+      "learning_rate": 9.190283400809717e-06,
+      "loss": 0.1894,
       "step": 200
     },
     {
       "epoch": 1.21,
-      "grad_norm": 4.295170783996582,
-      "learning_rate": 7.587044534412956e-06,
-      "loss": 0.7549,
+      "grad_norm": 0.05324831232428551,
+      "learning_rate": 8.785425101214575e-06,
+      "loss": 0.1251,
       "step": 300
     },
     {
       "epoch": 1.62,
-      "grad_norm": 2.47021746635437,
-      "learning_rate": 6.777327935222673e-06,
-      "loss": 0.545,
+      "grad_norm": 1.582890510559082,
+      "learning_rate": 8.380566801619434e-06,
+      "loss": 0.1728,
       "step": 400
     },
     {
       "epoch": 2.02,
-      "grad_norm": 4.9068498611450195,
-      "learning_rate": 5.9676113360323896e-06,
-      "loss": 0.4039,
+      "grad_norm": 9.203145027160645,
+      "learning_rate": 7.979757085020243e-06,
+      "loss": 0.1792,
       "step": 500
     },
     {
       "epoch": 2.02,
-      "eval_accuracy": 0.8949494957923889,
-      "eval_loss": 0.3197959065437317,
-      "eval_runtime": 43.5773,
-      "eval_samples_per_second": 11.359,
-      "eval_steps_per_second": 2.846,
+      "eval_accuracy": 0.931313157081604,
+      "eval_loss": 0.29975295066833496,
+      "eval_runtime": 23.1762,
+      "eval_samples_per_second": 21.358,
+      "eval_steps_per_second": 5.35,
       "step": 500
     },
     {
       "epoch": 2.42,
-      "grad_norm": 10.178821563720703,
-      "learning_rate": 5.157894736842106e-06,
-      "loss": 0.3572,
+      "grad_norm": 199.04556274414062,
+      "learning_rate": 7.574898785425102e-06,
+      "loss": 0.101,
       "step": 600
     },
     {
       "epoch": 2.83,
-      "grad_norm": 22.03152847290039,
-      "learning_rate": 4.348178137651822e-06,
-      "loss": 0.3237,
+      "grad_norm": 0.016058048233389854,
+      "learning_rate": 7.17004048582996e-06,
+      "loss": 0.1137,
       "step": 700
     },
     {
       "epoch": 3.23,
-      "grad_norm": 4.856521129608154,
-      "learning_rate": 3.538461538461539e-06,
-      "loss": 0.2635,
+      "grad_norm": 0.04301603510975838,
+      "learning_rate": 6.765182186234819e-06,
+      "loss": 0.1222,
       "step": 800
     },
     {
       "epoch": 3.64,
-      "grad_norm": 4.2341766357421875,
-      "learning_rate": 2.7368421052631583e-06,
-      "loss": 0.2138,
+      "grad_norm": 0.014038638211786747,
+      "learning_rate": 6.360323886639676e-06,
+      "loss": 0.1113,
       "step": 900
     },
     {
       "epoch": 4.04,
-      "grad_norm": 0.4156922996044159,
-      "learning_rate": 1.9271255060728746e-06,
-      "loss": 0.2195,
+      "grad_norm": 0.02426515519618988,
+      "learning_rate": 5.955465587044534e-06,
+      "loss": 0.1097,
       "step": 1000
     },
     {
       "epoch": 4.04,
-      "eval_accuracy": 0.9333333373069763,
-      "eval_loss": 0.2362980842590332,
-      "eval_runtime": 43.7442,
-      "eval_samples_per_second": 11.316,
-      "eval_steps_per_second": 2.835,
+      "eval_accuracy": 0.9474747180938721,
+      "eval_loss": 0.32810476422309875,
+      "eval_runtime": 23.0468,
+      "eval_samples_per_second": 21.478,
+      "eval_steps_per_second": 5.38,
       "step": 1000
     },
     {
       "epoch": 4.44,
-      "grad_norm": 5.4564337730407715,
-      "learning_rate": 1.1174089068825912e-06,
-      "loss": 0.2312,
+      "grad_norm": 10.28210735321045,
+      "learning_rate": 5.550607287449393e-06,
+      "loss": 0.1292,
       "step": 1100
     },
     {
       "epoch": 4.85,
-      "grad_norm": 2.0700230598449707,
-      "learning_rate": 3.0769230769230774e-07,
-      "loss": 0.182,
+      "grad_norm": 0.060118239372968674,
+      "learning_rate": 5.149797570850202e-06,
+      "loss": 0.0787,
       "step": 1200
     },
     {
-      "epoch": 4.99,
-      "step": 1235,
-      "total_flos": 1.053348926143392e+18,
-      "train_loss": 0.5674189640925481,
-      "train_runtime": 1664.3033,
-      "train_samples_per_second": 5.945,
-      "train_steps_per_second": 0.742
+      "epoch": 5.25,
+      "grad_norm": 0.5975450277328491,
+      "learning_rate": 4.744939271255061e-06,
+      "loss": 0.1137,
+      "step": 1300
+    },
+    {
+      "epoch": 5.66,
+      "grad_norm": 7.804176330566406,
+      "learning_rate": 4.34008097165992e-06,
+      "loss": 0.0449,
+      "step": 1400
+    },
+    {
+      "epoch": 6.06,
+      "grad_norm": 0.010911904275417328,
+      "learning_rate": 3.935222672064778e-06,
+      "loss": 0.0557,
+      "step": 1500
+    },
+    {
+      "epoch": 6.06,
+      "eval_accuracy": 0.965656578540802,
+      "eval_loss": 0.24546462297439575,
+      "eval_runtime": 23.0063,
+      "eval_samples_per_second": 21.516,
+      "eval_steps_per_second": 5.39,
+      "step": 1500
+    },
+    {
+      "epoch": 6.46,
+      "grad_norm": 0.11927049607038498,
+      "learning_rate": 3.5303643724696358e-06,
+      "loss": 0.0404,
+      "step": 1600
+    },
+    {
+      "epoch": 6.87,
+      "grad_norm": 0.0146423215046525,
+      "learning_rate": 3.125506072874494e-06,
+      "loss": 0.0785,
+      "step": 1700
+    },
+    {
+      "epoch": 7.27,
+      "grad_norm": 0.8568518757820129,
+      "learning_rate": 2.7206477732793525e-06,
+      "loss": 0.0589,
+      "step": 1800
+    },
+    {
+      "epoch": 7.68,
+      "grad_norm": 0.11240042001008987,
+      "learning_rate": 2.3157894736842105e-06,
+      "loss": 0.0688,
+      "step": 1900
+    },
+    {
+      "epoch": 8.08,
+      "grad_norm": 3.66904354095459,
+      "learning_rate": 1.910931174089069e-06,
+      "loss": 0.0535,
+      "step": 2000
+    },
+    {
+      "epoch": 8.08,
+      "eval_accuracy": 0.965656578540802,
+      "eval_loss": 0.22684885561466217,
+      "eval_runtime": 23.2645,
+      "eval_samples_per_second": 21.277,
+      "eval_steps_per_second": 5.33,
+      "step": 2000
+    },
+    {
+      "epoch": 8.48,
+      "grad_norm": 0.009745566174387932,
+      "learning_rate": 1.5060728744939273e-06,
+      "loss": 0.0542,
+      "step": 2100
+    },
+    {
+      "epoch": 8.89,
+      "grad_norm": 2.3708832263946533,
+      "learning_rate": 1.1012145748987855e-06,
+      "loss": 0.0506,
+      "step": 2200
+    },
+    {
+      "epoch": 9.29,
+      "grad_norm": 2.8488991260528564,
+      "learning_rate": 6.963562753036437e-07,
+      "loss": 0.0603,
+      "step": 2300
+    },
+    {
+      "epoch": 9.7,
+      "grad_norm": 0.012795313261449337,
+      "learning_rate": 2.9149797570850203e-07,
+      "loss": 0.04,
+      "step": 2400
+    },
+    {
+      "epoch": 9.98,
+      "step": 2470,
+      "total_flos": 6.32707029472752e+17,
+      "train_loss": 0.09667120742411749,
+      "train_runtime": 1319.7559,
+      "train_samples_per_second": 14.995,
+      "train_steps_per_second": 1.872
     }
   ],
   "logging_steps": 100,
-  "max_steps": 1235,
+  "max_steps": 2470,
   "num_input_tokens_seen": 0,
-  "num_train_epochs": 5,
+  "num_train_epochs": 10,
   "save_steps": 500,
-  "total_flos": 1.053348926143392e+18,
+  "total_flos": 6.32707029472752e+17,
   "train_batch_size": 4,
   "trial_name": null,
   "trial_params": null

training_args.bin

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:6aad4c2aa7f6067110d61ed613d6e4dbae9214f7578b02e94a2191162bee5aa5
+oid sha256:3bcd4dca7eb31ef8d31bc803bb33fd4e0608643c2c7770e14f65d990a3939b46
 size 4984