End of training

README.md

@@ -1,6 +1,6 @@
 ---
 license: apache-2.0
-base_model: ehcalabres/wav2vec2-lg-xlsr-en-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 [ehcalabres/wav2vec2-lg-xlsr-en-speech-emotion-recognition](https://huggingface.co/ehcalabres/wav2vec2-lg-xlsr-en-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: 0.0873
-- Accuracy: 0.9758
+- Loss: 0.2847
+- Accuracy: 0.9394
 
 ## Model description
 
@@ -50,15 +50,13 @@ The following hyperparameters were used during training:
 
 ### Training results
 
-| Training Loss | Epoch | Step | Validation Loss | Accuracy |
-|:-------------:|:-----:|:----:|:---------------:|:--------:|
-| 0.2538        | 2.02  | 500  | 0.2764          | 0.9071   |
-| 0.0805        | 4.04  | 1000 | 0.1722          | 0.9535   |
-| 0.0505        | 6.06  | 1500 | 0.1142          | 0.9778   |
-| 0.0466        | 8.08  | 2000 | 0.1177          | 0.9737   |
-| 0.0269        | 10.1  | 2500 | 0.1183          | 0.9737   |
-| 0.025         | 12.12 | 3000 | 0.0873          | 0.9758   |
-| 0.0323        | 14.14 | 3500 | 0.0920          | 0.9798   |
+| Training Loss | Epoch   | Step | Validation Loss | Accuracy |
+|:-------------:|:-------:|:----:|:---------------:|:--------:|
+| 0.4293        | 2.2422  | 500  | 0.4459          | 0.8788   |
+| 0.2063        | 4.4843  | 1000 | 0.2847          | 0.9394   |
+| 0.1392        | 6.7265  | 1500 | 0.3088          | 0.9394   |
+| 0.1434        | 8.9686  | 2000 | 0.3495          | 0.9444   |
+| 0.097         | 11.2108 | 2500 | 0.3493          | 0.9394   |
 
 
 ### Framework versions

all_results.json

@@ -1,9 +1,14 @@
 {
-    "epoch": 14.97,
-    "total_flos": 3.163398064220592e+18,
-    "train_loss": 0.12699281216954733,
-    "train_runtime": 5204.2292,
-    "train_samples": 1979,
-    "train_samples_per_second": 5.704,
-    "train_steps_per_second": 0.712
+    "epoch": 11.210762331838565,
+    "eval_accuracy": 0.939393937587738,
+    "eval_loss": 0.28471633791923523,
+    "eval_runtime": 9.1496,
+    "eval_samples_per_second": 21.64,
+    "eval_steps_per_second": 5.465,
+    "total_flos": 6.392309759902944e+17,
+    "train_loss": 0.27951997299194337,
+    "train_runtime": 1304.0962,
+    "train_samples": 1781,
+    "train_samples_per_second": 20.485,
+    "train_steps_per_second": 2.565
 }

config.json

@@ -1,20 +1,14 @@
 {
-  "_name_or_path": "ehcalabres/wav2vec2-lg-xlsr-en-speech-emotion-recognition",
-  "activation_dropout": 0.05,
-  "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": 0.05,
-  "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": 0.05,
-  "hidden_size": 1024,
+  "hidden_dropout": 0.1,
+  "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,
@@ -74,60 +69,25 @@
     "05Angry": 4
   },
   "layer_norm_eps": 1e-05,
-  "layerdrop": 0.05,
-  "mask_channel_length": 10,
-  "mask_channel_min_space": 1,
-  "mask_channel_other": 0.0,
-  "mask_channel_prob": 0.0,
-  "mask_channel_selection": "static",
+  "layerdrop": 0.1,
   "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,
+  "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": 33,
-  "xvector_output_dim": 512
+  "vocab_size": 32
 }

eval_results.json

@@ -0,0 +1,8 @@
+{
+    "epoch": 11.210762331838565,
+    "eval_accuracy": 0.939393937587738,
+    "eval_loss": 0.28471633791923523,
+    "eval_runtime": 9.1496,
+    "eval_samples_per_second": 21.64,
+    "eval_steps_per_second": 5.465
+}

model.safetensors

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:36ed8619730ada58c91dc801f90d035693a43655ce5856d2dc3968d3fc2e881d
-size 1266026604
+oid sha256:46650e61181300624f9711b1e316f3a331d5d40bf61280f4ee96cf2619f293c1
+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/Apr17_14-56-56_43c303507aad/events.out.tfevents.1713365817.43c303507aad.34.0

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:ec7a64a78f16680c6bfa14aed11449ee92c9f25fc469e1e854defd6a5f97563a
+size 17750

runs/Apr17_15-02-07_43c303507aad/events.out.tfevents.1713366132.43c303507aad.34.1

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:e37a581a621fe12664f2ed939269df9eabbf3a552aa04e41d4feb0040882ac75
+size 13179

runs/Apr17_15-02-07_43c303507aad/events.out.tfevents.1713367446.43c303507aad.34.2

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:bcd4d3b778bd2f338917f4e897fc13537ea4b098b5eadc28dc4bc957ca9b8b06
+size 734

train_results.json

@@ -1,9 +1,9 @@
 {
-    "epoch": 14.97,
-    "total_flos": 3.163398064220592e+18,
-    "train_loss": 0.12699281216954733,
-    "train_runtime": 5204.2292,
-    "train_samples": 1979,
-    "train_samples_per_second": 5.704,
-    "train_steps_per_second": 0.712
+    "epoch": 11.210762331838565,
+    "total_flos": 6.392309759902944e+17,
+    "train_loss": 0.27951997299194337,
+    "train_runtime": 1304.0962,
+    "train_samples": 1781,
+    "train_samples_per_second": 20.485,
+    "train_steps_per_second": 2.565
 }

trainer_state.json

@@ -1,351 +1,258 @@
 {
-  "best_metric": 0.08732648193836212,
-  "best_model_checkpoint": "/kaggle/working/ckpts/checkpoint-3000",
-  "epoch": 14.969696969696969,
+  "best_metric": 0.28471633791923523,
+  "best_model_checkpoint": "/kaggle/working/ckpts/checkpoint-1000",
+  "epoch": 11.210762331838565,
   "eval_steps": 500,
-  "global_step": 3705,
+  "global_step": 2500,
   "is_hyper_param_search": false,
   "is_local_process_zero": true,
   "is_world_process_zero": true,
   "log_history": [
     {
-      "epoch": 0.4,
-      "grad_norm": 4.305359363555908,
-      "learning_rate": 9.730094466936572e-06,
-      "loss": 1.2225,
+      "epoch": 0.4484304932735426,
+      "grad_norm": 5.388023376464844,
+      "learning_rate": 9.701046337817639e-06,
+      "loss": 1.1493,
       "step": 100
     },
     {
-      "epoch": 0.81,
-      "grad_norm": 4.125752925872803,
-      "learning_rate": 9.460188933873145e-06,
-      "loss": 0.6537,
+      "epoch": 0.8968609865470852,
+      "grad_norm": 3.5460991859436035,
+      "learning_rate": 9.4050822122571e-06,
+      "loss": 0.8211,
       "step": 200
     },
     {
-      "epoch": 1.21,
-      "grad_norm": 1.4285308122634888,
-      "learning_rate": 9.190283400809717e-06,
-      "loss": 0.3636,
+      "epoch": 1.3452914798206277,
+      "grad_norm": 10.436558723449707,
+      "learning_rate": 9.109118086696563e-06,
+      "loss": 0.6329,
       "step": 300
     },
     {
-      "epoch": 1.62,
-      "grad_norm": 0.3685835301876068,
-      "learning_rate": 8.92037786774629e-06,
-      "loss": 0.2794,
+      "epoch": 1.7937219730941703,
+      "grad_norm": 3.8405914306640625,
+      "learning_rate": 8.810164424514201e-06,
+      "loss": 0.4802,
       "step": 400
     },
     {
-      "epoch": 2.02,
-      "grad_norm": 9.573077201843262,
-      "learning_rate": 8.653171390013496e-06,
-      "loss": 0.2538,
+      "epoch": 2.242152466367713,
+      "grad_norm": 18.780351638793945,
+      "learning_rate": 8.51121076233184e-06,
+      "loss": 0.4293,
       "step": 500
     },
     {
-      "epoch": 2.02,
-      "eval_accuracy": 0.9070706963539124,
-      "eval_loss": 0.2763901352882385,
-      "eval_runtime": 44.7688,
-      "eval_samples_per_second": 11.057,
-      "eval_steps_per_second": 2.77,
+      "epoch": 2.242152466367713,
+      "eval_accuracy": 0.8787878751754761,
+      "eval_loss": 0.44592228531837463,
+      "eval_runtime": 9.3047,
+      "eval_samples_per_second": 21.28,
+      "eval_steps_per_second": 5.374,
       "step": 500
     },
     {
-      "epoch": 2.42,
-      "grad_norm": 23.30544090270996,
-      "learning_rate": 8.385964912280704e-06,
-      "loss": 0.1608,
+      "epoch": 2.6905829596412554,
+      "grad_norm": 1.012455940246582,
+      "learning_rate": 8.212257100149478e-06,
+      "loss": 0.296,
       "step": 600
     },
     {
-      "epoch": 2.83,
-      "grad_norm": 0.28133082389831543,
-      "learning_rate": 8.116059379217275e-06,
-      "loss": 0.149,
+      "epoch": 3.1390134529147984,
+      "grad_norm": 15.63992691040039,
+      "learning_rate": 7.913303437967116e-06,
+      "loss": 0.3031,
       "step": 700
     },
     {
-      "epoch": 3.23,
-      "grad_norm": 20.600244522094727,
-      "learning_rate": 7.846153846153847e-06,
-      "loss": 0.1573,
+      "epoch": 3.587443946188341,
+      "grad_norm": 11.77250862121582,
+      "learning_rate": 7.617339312406578e-06,
+      "loss": 0.2514,
       "step": 800
     },
     {
-      "epoch": 3.64,
-      "grad_norm": 5.6504645347595215,
-      "learning_rate": 7.576248313090419e-06,
-      "loss": 0.1273,
+      "epoch": 4.0358744394618835,
+      "grad_norm": 41.904685974121094,
+      "learning_rate": 7.318385650224216e-06,
+      "loss": 0.2968,
       "step": 900
     },
     {
-      "epoch": 4.04,
-      "grad_norm": 0.06874903291463852,
-      "learning_rate": 7.306342780026991e-06,
-      "loss": 0.0805,
+      "epoch": 4.484304932735426,
+      "grad_norm": 3.609229326248169,
+      "learning_rate": 7.019431988041854e-06,
+      "loss": 0.2063,
       "step": 1000
     },
     {
-      "epoch": 4.04,
-      "eval_accuracy": 0.9535353779792786,
-      "eval_loss": 0.1721607893705368,
-      "eval_runtime": 45.1822,
-      "eval_samples_per_second": 10.956,
-      "eval_steps_per_second": 2.744,
+      "epoch": 4.484304932735426,
+      "eval_accuracy": 0.939393937587738,
+      "eval_loss": 0.28471633791923523,
+      "eval_runtime": 9.1772,
+      "eval_samples_per_second": 21.575,
+      "eval_steps_per_second": 5.448,
       "step": 1000
     },
     {
-      "epoch": 4.44,
-      "grad_norm": 2.5783190727233887,
-      "learning_rate": 7.036437246963563e-06,
-      "loss": 0.124,
+      "epoch": 4.932735426008969,
+      "grad_norm": 40.831077575683594,
+      "learning_rate": 6.720478325859492e-06,
+      "loss": 0.189,
       "step": 1100
     },
     {
-      "epoch": 4.85,
-      "grad_norm": 0.08274731785058975,
-      "learning_rate": 6.766531713900135e-06,
-      "loss": 0.0919,
+      "epoch": 5.381165919282511,
+      "grad_norm": 6.550039768218994,
+      "learning_rate": 6.421524663677131e-06,
+      "loss": 0.1695,
       "step": 1200
     },
     {
-      "epoch": 5.25,
-      "grad_norm": 0.7308838367462158,
-      "learning_rate": 6.496626180836708e-06,
-      "loss": 0.0882,
+      "epoch": 5.829596412556054,
+      "grad_norm": 0.8909838795661926,
+      "learning_rate": 6.1225710014947695e-06,
+      "loss": 0.1986,
       "step": 1300
     },
     {
-      "epoch": 5.66,
-      "grad_norm": 0.09978947788476944,
-      "learning_rate": 6.22672064777328e-06,
-      "loss": 0.0536,
+      "epoch": 6.278026905829597,
+      "grad_norm": 1.107909917831421,
+      "learning_rate": 5.823617339312408e-06,
+      "loss": 0.1881,
       "step": 1400
     },
     {
-      "epoch": 6.06,
-      "grad_norm": 0.02113749459385872,
-      "learning_rate": 5.956815114709852e-06,
-      "loss": 0.0505,
+      "epoch": 6.726457399103139,
+      "grad_norm": 0.359598308801651,
+      "learning_rate": 5.524663677130046e-06,
+      "loss": 0.1392,
       "step": 1500
     },
     {
-      "epoch": 6.06,
-      "eval_accuracy": 0.9777777791023254,
-      "eval_loss": 0.11419288069009781,
-      "eval_runtime": 45.0408,
-      "eval_samples_per_second": 10.99,
-      "eval_steps_per_second": 2.753,
+      "epoch": 6.726457399103139,
+      "eval_accuracy": 0.939393937587738,
+      "eval_loss": 0.3088470995426178,
+      "eval_runtime": 9.2392,
+      "eval_samples_per_second": 21.43,
+      "eval_steps_per_second": 5.412,
       "step": 1500
     },
     {
-      "epoch": 6.46,
-      "grad_norm": 0.04003220424056053,
-      "learning_rate": 5.686909581646424e-06,
-      "loss": 0.0283,
+      "epoch": 7.174887892376682,
+      "grad_norm": 0.030851535499095917,
+      "learning_rate": 5.225710014947683e-06,
+      "loss": 0.1173,
       "step": 1600
     },
     {
-      "epoch": 6.87,
-      "grad_norm": 0.01778920739889145,
-      "learning_rate": 5.417004048582997e-06,
-      "loss": 0.0862,
+      "epoch": 7.623318385650224,
+      "grad_norm": 0.03430308774113655,
+      "learning_rate": 4.9267563527653214e-06,
+      "loss": 0.1522,
       "step": 1700
     },
     {
-      "epoch": 7.27,
-      "grad_norm": 2.5195560455322266,
-      "learning_rate": 5.147098515519568e-06,
-      "loss": 0.0623,
+      "epoch": 8.071748878923767,
+      "grad_norm": 4.6850409507751465,
+      "learning_rate": 4.62780269058296e-06,
+      "loss": 0.1632,
       "step": 1800
     },
     {
-      "epoch": 7.68,
-      "grad_norm": 0.016020264476537704,
-      "learning_rate": 4.877192982456141e-06,
-      "loss": 0.0622,
+      "epoch": 8.52017937219731,
+      "grad_norm": 0.036871179938316345,
+      "learning_rate": 4.328849028400599e-06,
+      "loss": 0.1147,
       "step": 1900
     },
     {
-      "epoch": 8.08,
-      "grad_norm": 0.1588612198829651,
-      "learning_rate": 4.607287449392713e-06,
-      "loss": 0.0466,
+      "epoch": 8.968609865470851,
+      "grad_norm": 73.54190063476562,
+      "learning_rate": 4.029895366218237e-06,
+      "loss": 0.1434,
       "step": 2000
     },
     {
-      "epoch": 8.08,
-      "eval_accuracy": 0.973737359046936,
-      "eval_loss": 0.11773889511823654,
-      "eval_runtime": 44.6026,
-      "eval_samples_per_second": 11.098,
-      "eval_steps_per_second": 2.78,
+      "epoch": 8.968609865470851,
+      "eval_accuracy": 0.9444444179534912,
+      "eval_loss": 0.349543035030365,
+      "eval_runtime": 9.3502,
+      "eval_samples_per_second": 21.176,
+      "eval_steps_per_second": 5.347,
       "step": 2000
     },
     {
-      "epoch": 8.48,
-      "grad_norm": 0.017447825521230698,
-      "learning_rate": 4.337381916329285e-06,
-      "loss": 0.0372,
+      "epoch": 9.417040358744394,
+      "grad_norm": 0.03475332260131836,
+      "learning_rate": 3.7309417040358746e-06,
+      "loss": 0.1444,
       "step": 2100
     },
     {
-      "epoch": 8.89,
-      "grad_norm": 0.07465946674346924,
-      "learning_rate": 4.067476383265857e-06,
-      "loss": 0.0338,
+      "epoch": 9.865470852017937,
+      "grad_norm": 0.26133427023887634,
+      "learning_rate": 3.431988041853513e-06,
+      "loss": 0.101,
       "step": 2200
     },
     {
-      "epoch": 9.29,
-      "grad_norm": 1.1038620471954346,
-      "learning_rate": 3.800269905533064e-06,
-      "loss": 0.0458,
+      "epoch": 10.31390134529148,
+      "grad_norm": 0.03596407175064087,
+      "learning_rate": 3.133034379671151e-06,
+      "loss": 0.0849,
       "step": 2300
     },
     {
-      "epoch": 9.7,
-      "grad_norm": 0.033347614109516144,
-      "learning_rate": 3.5303643724696358e-06,
-      "loss": 0.0491,
+      "epoch": 10.762331838565022,
+      "grad_norm": 0.03674998879432678,
+      "learning_rate": 2.8340807174887896e-06,
+      "loss": 0.119,
       "step": 2400
     },
     {
-      "epoch": 10.1,
-      "grad_norm": 0.010389924980700016,
-      "learning_rate": 3.260458839406208e-06,
-      "loss": 0.0269,
+      "epoch": 11.210762331838565,
+      "grad_norm": 0.2922690510749817,
+      "learning_rate": 2.538116591928251e-06,
+      "loss": 0.097,
       "step": 2500
     },
     {
-      "epoch": 10.1,
-      "eval_accuracy": 0.973737359046936,
-      "eval_loss": 0.11825177818536758,
-      "eval_runtime": 44.732,
-      "eval_samples_per_second": 11.066,
-      "eval_steps_per_second": 2.772,
+      "epoch": 11.210762331838565,
+      "eval_accuracy": 0.939393937587738,
+      "eval_loss": 0.34934303164482117,
+      "eval_runtime": 9.3163,
+      "eval_samples_per_second": 21.253,
+      "eval_steps_per_second": 5.367,
       "step": 2500
     },
     {
-      "epoch": 10.51,
-      "grad_norm": 0.014653917402029037,
-      "learning_rate": 2.9905533063427807e-06,
-      "loss": 0.0517,
-      "step": 2600
+      "epoch": 11.210762331838565,
+      "step": 2500,
+      "total_flos": 6.392309759902944e+17,
+      "train_loss": 0.27951997299194337,
+      "train_runtime": 1304.0962,
+      "train_samples_per_second": 20.485,
+      "train_steps_per_second": 2.565
     },
     {
-      "epoch": 10.91,
-      "grad_norm": 0.012147756293416023,
-      "learning_rate": 2.7206477732793525e-06,
-      "loss": 0.0424,
-      "step": 2700
-    },
-    {
-      "epoch": 11.31,
-      "grad_norm": 0.018718041479587555,
-      "learning_rate": 2.4507422402159244e-06,
-      "loss": 0.0216,
-      "step": 2800
-    },
-    {
-      "epoch": 11.72,
-      "grad_norm": 0.012194407172501087,
-      "learning_rate": 2.180836707152497e-06,
-      "loss": 0.0219,
-      "step": 2900
-    },
-    {
-      "epoch": 12.12,
-      "grad_norm": 0.06089940294623375,
-      "learning_rate": 1.910931174089069e-06,
-      "loss": 0.025,
-      "step": 3000
-    },
-    {
-      "epoch": 12.12,
-      "eval_accuracy": 0.9757575988769531,
-      "eval_loss": 0.08732648193836212,
-      "eval_runtime": 44.5561,
-      "eval_samples_per_second": 11.11,
-      "eval_steps_per_second": 2.783,
-      "step": 3000
-    },
-    {
-      "epoch": 12.53,
-      "grad_norm": 0.07689320296049118,
-      "learning_rate": 1.6410256410256412e-06,
-      "loss": 0.0398,
-      "step": 3100
-    },
-    {
-      "epoch": 12.93,
-      "grad_norm": 0.35102277994155884,
-      "learning_rate": 1.3711201079622132e-06,
-      "loss": 0.0289,
-      "step": 3200
-    },
-    {
-      "epoch": 13.33,
-      "grad_norm": 0.007283875718712807,
-      "learning_rate": 1.1012145748987855e-06,
-      "loss": 0.0169,
-      "step": 3300
-    },
-    {
-      "epoch": 13.74,
-      "grad_norm": 0.00747378496453166,
-      "learning_rate": 8.313090418353577e-07,
-      "loss": 0.0391,
-      "step": 3400
-    },
-    {
-      "epoch": 14.14,
-      "grad_norm": 0.0248583871871233,
-      "learning_rate": 5.614035087719299e-07,
-      "loss": 0.0323,
-      "step": 3500
-    },
-    {
-      "epoch": 14.14,
-      "eval_accuracy": 0.9797979593276978,
-      "eval_loss": 0.09203575551509857,
-      "eval_runtime": 46.8049,
-      "eval_samples_per_second": 10.576,
-      "eval_steps_per_second": 2.649,
-      "step": 3500
-    },
-    {
-      "epoch": 14.55,
-      "grad_norm": 0.1301543265581131,
-      "learning_rate": 2.9149797570850203e-07,
-      "loss": 0.0252,
-      "step": 3600
-    },
-    {
-      "epoch": 14.95,
-      "grad_norm": 0.09905605018138885,
-      "learning_rate": 2.1592442645074224e-08,
-      "loss": 0.0146,
-      "step": 3700
-    },
-    {
-      "epoch": 14.97,
-      "step": 3705,
-      "total_flos": 3.163398064220592e+18,
-      "train_loss": 0.12699281216954733,
-      "train_runtime": 5204.2292,
-      "train_samples_per_second": 5.704,
-      "train_steps_per_second": 0.712
+      "epoch": 11.210762331838565,
+      "eval_accuracy": 0.939393937587738,
+      "eval_loss": 0.28471633791923523,
+      "eval_runtime": 9.1496,
+      "eval_samples_per_second": 21.64,
+      "eval_steps_per_second": 5.465,
+      "step": 2500
     }
   ],
   "logging_steps": 100,
-  "max_steps": 3705,
+  "max_steps": 3345,
   "num_input_tokens_seen": 0,
   "num_train_epochs": 15,
   "save_steps": 500,
-  "total_flos": 3.163398064220592e+18,
+  "total_flos": 6.392309759902944e+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:ac6c83dae5ee0fc426b289b9abef79c6451dd3a371db69497611b3ba3258e6ec
+oid sha256:11653a441445d7e0f57f6fc96f2220c1565b1734c352c0402119f682a0beb602
 size 4984