Upload batch 9/24: mrpc/mBERT/gated_multi_branch/seed_44 ... mrpc/mBERT/hf_sequence_classifier/seed_44

mbert_rtx6000_metrics/result_ablation_mbert/mrpc/mBERT/gated_multi_branch/seed_44/all_results.json

@@ -0,0 +1,17 @@
+{
+    "epoch": 3.0,
+    "eval_accuracy": 0.8602941176470589,
+    "eval_combined_score": 0.8787957074721782,
+    "eval_f1": 0.8972972972972973,
+    "eval_loss": 0.34184572100639343,
+    "eval_runtime": 1.3702,
+    "eval_samples": 408,
+    "eval_samples_per_second": 297.757,
+    "eval_steps_per_second": 3.649,
+    "total_flos": 649699269630048.0,
+    "train_loss": 0.385329819344855,
+    "train_runtime": 33.4079,
+    "train_samples": 3668,
+    "train_samples_per_second": 329.383,
+    "train_steps_per_second": 6.915
+}

mbert_rtx6000_metrics/result_ablation_mbert/mrpc/mBERT/gated_multi_branch/seed_44/eval_results.json

@@ -0,0 +1,11 @@
+{
+    "epoch": 3.0,
+    "eval_accuracy": 0.8602941176470589,
+    "eval_combined_score": 0.8787957074721782,
+    "eval_f1": 0.8972972972972973,
+    "eval_loss": 0.34184572100639343,
+    "eval_runtime": 1.3702,
+    "eval_samples": 408,
+    "eval_samples_per_second": 297.757,
+    "eval_steps_per_second": 3.649
+}

mbert_rtx6000_metrics/result_ablation_mbert/mrpc/mBERT/gated_multi_branch/seed_44/run_command.txt

@@ -0,0 +1 @@
+/usr/local/bin/python /workspace/run_glue_MHA_gated.py --model_name_or_path bert-base-multilingual-cased --dataset_name tmnam20/ViGLUE --task_name mrpc --max_seq_length 128 --do_train --do_eval --evaluation_strategy epoch --save_strategy no --logging_strategy steps --logging_steps 25 --report_to none --num_train_epochs 3 --learning_rate 3e-05 --per_device_train_batch_size 48 --per_device_eval_batch_size 96 --gradient_accumulation_steps 1 --dataloader_num_workers 1 --seed 44 --output_dir result_ablation_mbert/mrpc/mBERT/gated_multi_branch/seed_44 --overwrite_output_dir --trust_remote_code true --bf16 --tf32 true --skip_model_save --use_mha true --pooling_strategy gated_multi_branch --n_heads 4 --head_dim 32 --classifier_hidden_dim 512 --head_dropout 0.05

mbert_rtx6000_metrics/result_ablation_mbert/mrpc/mBERT/gated_multi_branch/seed_44/train_results.json

@@ -0,0 +1,9 @@
+{
+    "epoch": 3.0,
+    "total_flos": 649699269630048.0,
+    "train_loss": 0.385329819344855,
+    "train_runtime": 33.4079,
+    "train_samples": 3668,
+    "train_samples_per_second": 329.383,
+    "train_steps_per_second": 6.915
+}

mbert_rtx6000_metrics/result_ablation_mbert/mrpc/mBERT/gated_multi_branch/seed_44/trainer_state.json

@@ -0,0 +1,138 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 3.0,
+  "eval_steps": 500,
+  "global_step": 231,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.3246753246753247,
+      "grad_norm": 1.9964197874069214,
+      "learning_rate": 2.6753246753246755e-05,
+      "loss": 0.5994,
+      "step": 25
+    },
+    {
+      "epoch": 0.6493506493506493,
+      "grad_norm": 2.559793710708618,
+      "learning_rate": 2.350649350649351e-05,
+      "loss": 0.5489,
+      "step": 50
+    },
+    {
+      "epoch": 0.974025974025974,
+      "grad_norm": 4.578395366668701,
+      "learning_rate": 2.025974025974026e-05,
+      "loss": 0.4804,
+      "step": 75
+    },
+    {
+      "epoch": 1.0,
+      "eval_accuracy": 0.8235294117647058,
+      "eval_combined_score": 0.8419972640218878,
+      "eval_f1": 0.8604651162790697,
+      "eval_loss": 0.41182783246040344,
+      "eval_runtime": 1.83,
+      "eval_samples_per_second": 222.95,
+      "eval_steps_per_second": 2.732,
+      "step": 77
+    },
+    {
+      "epoch": 1.2987012987012987,
+      "grad_norm": 4.026550769805908,
+      "learning_rate": 1.7012987012987013e-05,
+      "loss": 0.3802,
+      "step": 100
+    },
+    {
+      "epoch": 1.6233766233766234,
+      "grad_norm": 4.639058589935303,
+      "learning_rate": 1.3766233766233767e-05,
+      "loss": 0.376,
+      "step": 125
+    },
+    {
+      "epoch": 1.948051948051948,
+      "grad_norm": 4.616215705871582,
+      "learning_rate": 1.051948051948052e-05,
+      "loss": 0.3741,
+      "step": 150
+    },
+    {
+      "epoch": 2.0,
+      "eval_accuracy": 0.8676470588235294,
+      "eval_combined_score": 0.8862883181441591,
+      "eval_f1": 0.9049295774647887,
+      "eval_loss": 0.3325900733470917,
+      "eval_runtime": 1.3292,
+      "eval_samples_per_second": 306.941,
+      "eval_steps_per_second": 3.762,
+      "step": 154
+    },
+    {
+      "epoch": 2.2727272727272725,
+      "grad_norm": 7.209536075592041,
+      "learning_rate": 7.272727272727273e-06,
+      "loss": 0.2708,
+      "step": 175
+    },
+    {
+      "epoch": 2.5974025974025974,
+      "grad_norm": 7.712475299835205,
+      "learning_rate": 4.025974025974026e-06,
+      "loss": 0.2348,
+      "step": 200
+    },
+    {
+      "epoch": 2.9220779220779223,
+      "grad_norm": 4.1902971267700195,
+      "learning_rate": 7.792207792207793e-07,
+      "loss": 0.2353,
+      "step": 225
+    },
+    {
+      "epoch": 3.0,
+      "eval_accuracy": 0.8602941176470589,
+      "eval_combined_score": 0.8787957074721782,
+      "eval_f1": 0.8972972972972973,
+      "eval_loss": 0.34184572100639343,
+      "eval_runtime": 1.2953,
+      "eval_samples_per_second": 314.976,
+      "eval_steps_per_second": 3.86,
+      "step": 231
+    },
+    {
+      "epoch": 3.0,
+      "step": 231,
+      "total_flos": 649699269630048.0,
+      "train_loss": 0.385329819344855,
+      "train_runtime": 33.4079,
+      "train_samples_per_second": 329.383,
+      "train_steps_per_second": 6.915
+    }
+  ],
+  "logging_steps": 25,
+  "max_steps": 231,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 3,
+  "save_steps": 500,
+  "stateful_callbacks": {
+    "TrainerControl": {
+      "args": {
+        "should_epoch_stop": false,
+        "should_evaluate": false,
+        "should_log": false,
+        "should_save": false,
+        "should_training_stop": false
+      },
+      "attributes": {}
+    }
+  },
+  "total_flos": 649699269630048.0,
+  "train_batch_size": 48,
+  "trial_name": null,
+  "trial_params": null
+}

mbert_rtx6000_metrics/result_ablation_mbert/mrpc/mBERT/hf_sequence_classifier/seed_42/all_results.json

@@ -0,0 +1,17 @@
+{
+    "epoch": 3.0,
+    "eval_accuracy": 0.8161764705882353,
+    "eval_combined_score": 0.8382558330594807,
+    "eval_f1": 0.8603351955307262,
+    "eval_loss": 0.37590235471725464,
+    "eval_runtime": 1.8425,
+    "eval_samples": 408,
+    "eval_samples_per_second": 221.44,
+    "eval_steps_per_second": 2.714,
+    "total_flos": 627731589210960.0,
+    "train_loss": 0.44336734757278906,
+    "train_runtime": 41.466,
+    "train_samples": 3668,
+    "train_samples_per_second": 265.374,
+    "train_steps_per_second": 5.571
+}

mbert_rtx6000_metrics/result_ablation_mbert/mrpc/mBERT/hf_sequence_classifier/seed_42/eval_results.json

@@ -0,0 +1,11 @@
+{
+    "epoch": 3.0,
+    "eval_accuracy": 0.8161764705882353,
+    "eval_combined_score": 0.8382558330594807,
+    "eval_f1": 0.8603351955307262,
+    "eval_loss": 0.37590235471725464,
+    "eval_runtime": 1.8425,
+    "eval_samples": 408,
+    "eval_samples_per_second": 221.44,
+    "eval_steps_per_second": 2.714
+}

mbert_rtx6000_metrics/result_ablation_mbert/mrpc/mBERT/hf_sequence_classifier/seed_42/run_command.txt

@@ -0,0 +1 @@
+/usr/local/bin/python /workspace/run_glue_MHA_gated.py --model_name_or_path bert-base-multilingual-cased --dataset_name tmnam20/ViGLUE --task_name mrpc --max_seq_length 128 --do_train --do_eval --evaluation_strategy epoch --save_strategy no --logging_strategy steps --logging_steps 25 --report_to none --num_train_epochs 3 --learning_rate 3e-05 --per_device_train_batch_size 48 --per_device_eval_batch_size 96 --gradient_accumulation_steps 1 --dataloader_num_workers 1 --seed 42 --output_dir result_ablation_mbert/mrpc/mBERT/hf_sequence_classifier/seed_42 --overwrite_output_dir --trust_remote_code true --bf16 --tf32 true --skip_model_save

mbert_rtx6000_metrics/result_ablation_mbert/mrpc/mBERT/hf_sequence_classifier/seed_42/train_results.json

@@ -0,0 +1,9 @@
+{
+    "epoch": 3.0,
+    "total_flos": 627731589210960.0,
+    "train_loss": 0.44336734757278906,
+    "train_runtime": 41.466,
+    "train_samples": 3668,
+    "train_samples_per_second": 265.374,
+    "train_steps_per_second": 5.571
+}

mbert_rtx6000_metrics/result_ablation_mbert/mrpc/mBERT/hf_sequence_classifier/seed_42/trainer_state.json

@@ -0,0 +1,138 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 3.0,
+  "eval_steps": 500,
+  "global_step": 231,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.3246753246753247,
+      "grad_norm": 2.0890960693359375,
+      "learning_rate": 2.6753246753246755e-05,
+      "loss": 0.6061,
+      "step": 25
+    },
+    {
+      "epoch": 0.6493506493506493,
+      "grad_norm": 4.14626407623291,
+      "learning_rate": 2.350649350649351e-05,
+      "loss": 0.5293,
+      "step": 50
+    },
+    {
+      "epoch": 0.974025974025974,
+      "grad_norm": 3.3394346237182617,
+      "learning_rate": 2.025974025974026e-05,
+      "loss": 0.539,
+      "step": 75
+    },
+    {
+      "epoch": 1.0,
+      "eval_accuracy": 0.7794117647058824,
+      "eval_combined_score": 0.8115808823529411,
+      "eval_f1": 0.84375,
+      "eval_loss": 0.4522019922733307,
+      "eval_runtime": 1.7874,
+      "eval_samples_per_second": 228.259,
+      "eval_steps_per_second": 2.797,
+      "step": 77
+    },
+    {
+      "epoch": 1.2987012987012987,
+      "grad_norm": 5.786990165710449,
+      "learning_rate": 1.7012987012987013e-05,
+      "loss": 0.4253,
+      "step": 100
+    },
+    {
+      "epoch": 1.6233766233766234,
+      "grad_norm": 11.188650131225586,
+      "learning_rate": 1.3766233766233767e-05,
+      "loss": 0.4634,
+      "step": 125
+    },
+    {
+      "epoch": 1.948051948051948,
+      "grad_norm": 5.304947376251221,
+      "learning_rate": 1.051948051948052e-05,
+      "loss": 0.4306,
+      "step": 150
+    },
+    {
+      "epoch": 2.0,
+      "eval_accuracy": 0.8014705882352942,
+      "eval_combined_score": 0.8230001693575704,
+      "eval_f1": 0.8445297504798465,
+      "eval_loss": 0.4265172481536865,
+      "eval_runtime": 1.8415,
+      "eval_samples_per_second": 221.554,
+      "eval_steps_per_second": 2.715,
+      "step": 154
+    },
+    {
+      "epoch": 2.2727272727272725,
+      "grad_norm": 3.4806904792785645,
+      "learning_rate": 7.272727272727273e-06,
+      "loss": 0.379,
+      "step": 175
+    },
+    {
+      "epoch": 2.5974025974025974,
+      "grad_norm": 7.440636157989502,
+      "learning_rate": 4.025974025974026e-06,
+      "loss": 0.3246,
+      "step": 200
+    },
+    {
+      "epoch": 2.9220779220779223,
+      "grad_norm": 8.39614200592041,
+      "learning_rate": 7.792207792207793e-07,
+      "loss": 0.3212,
+      "step": 225
+    },
+    {
+      "epoch": 3.0,
+      "eval_accuracy": 0.8161764705882353,
+      "eval_combined_score": 0.8382558330594807,
+      "eval_f1": 0.8603351955307262,
+      "eval_loss": 0.37590235471725464,
+      "eval_runtime": 1.717,
+      "eval_samples_per_second": 237.628,
+      "eval_steps_per_second": 2.912,
+      "step": 231
+    },
+    {
+      "epoch": 3.0,
+      "step": 231,
+      "total_flos": 627731589210960.0,
+      "train_loss": 0.44336734757278906,
+      "train_runtime": 41.466,
+      "train_samples_per_second": 265.374,
+      "train_steps_per_second": 5.571
+    }
+  ],
+  "logging_steps": 25,
+  "max_steps": 231,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 3,
+  "save_steps": 500,
+  "stateful_callbacks": {
+    "TrainerControl": {
+      "args": {
+        "should_epoch_stop": false,
+        "should_evaluate": false,
+        "should_log": false,
+        "should_save": false,
+        "should_training_stop": false
+      },
+      "attributes": {}
+    }
+  },
+  "total_flos": 627731589210960.0,
+  "train_batch_size": 48,
+  "trial_name": null,
+  "trial_params": null
+}

mbert_rtx6000_metrics/result_ablation_mbert/mrpc/mBERT/hf_sequence_classifier/seed_43/all_results.json

@@ -0,0 +1,17 @@
+{
+    "epoch": 3.0,
+    "eval_accuracy": 0.8382352941176471,
+    "eval_combined_score": 0.859977862112587,
+    "eval_f1": 0.8817204301075269,
+    "eval_loss": 0.3466010093688965,
+    "eval_runtime": 1.5932,
+    "eval_samples": 408,
+    "eval_samples_per_second": 256.09,
+    "eval_steps_per_second": 3.138,
+    "total_flos": 625721256303600.0,
+    "train_loss": 0.41218561940378956,
+    "train_runtime": 20.9554,
+    "train_samples": 3668,
+    "train_samples_per_second": 525.114,
+    "train_steps_per_second": 11.023
+}

mbert_rtx6000_metrics/result_ablation_mbert/mrpc/mBERT/hf_sequence_classifier/seed_43/eval_results.json

@@ -0,0 +1,11 @@
+{
+    "epoch": 3.0,
+    "eval_accuracy": 0.8382352941176471,
+    "eval_combined_score": 0.859977862112587,
+    "eval_f1": 0.8817204301075269,
+    "eval_loss": 0.3466010093688965,
+    "eval_runtime": 1.5932,
+    "eval_samples": 408,
+    "eval_samples_per_second": 256.09,
+    "eval_steps_per_second": 3.138
+}

mbert_rtx6000_metrics/result_ablation_mbert/mrpc/mBERT/hf_sequence_classifier/seed_43/run_command.txt

@@ -0,0 +1 @@
+/usr/local/bin/python /workspace/run_glue_MHA_gated.py --model_name_or_path bert-base-multilingual-cased --dataset_name tmnam20/ViGLUE --task_name mrpc --max_seq_length 128 --do_train --do_eval --evaluation_strategy epoch --save_strategy no --logging_strategy steps --logging_steps 25 --report_to none --num_train_epochs 3 --learning_rate 3e-05 --per_device_train_batch_size 48 --per_device_eval_batch_size 96 --gradient_accumulation_steps 1 --dataloader_num_workers 1 --seed 43 --output_dir result_ablation_mbert/mrpc/mBERT/hf_sequence_classifier/seed_43 --overwrite_output_dir --trust_remote_code true --bf16 --tf32 true --skip_model_save

mbert_rtx6000_metrics/result_ablation_mbert/mrpc/mBERT/hf_sequence_classifier/seed_43/train_results.json

@@ -0,0 +1,9 @@
+{
+    "epoch": 3.0,
+    "total_flos": 625721256303600.0,
+    "train_loss": 0.41218561940378956,
+    "train_runtime": 20.9554,
+    "train_samples": 3668,
+    "train_samples_per_second": 525.114,
+    "train_steps_per_second": 11.023
+}

mbert_rtx6000_metrics/result_ablation_mbert/mrpc/mBERT/hf_sequence_classifier/seed_43/trainer_state.json

@@ -0,0 +1,138 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 3.0,
+  "eval_steps": 500,
+  "global_step": 231,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.3246753246753247,
+      "grad_norm": 3.668508291244507,
+      "learning_rate": 2.6753246753246755e-05,
+      "loss": 0.586,
+      "step": 25
+    },
+    {
+      "epoch": 0.6493506493506493,
+      "grad_norm": 5.74367618560791,
+      "learning_rate": 2.350649350649351e-05,
+      "loss": 0.5549,
+      "step": 50
+    },
+    {
+      "epoch": 0.974025974025974,
+      "grad_norm": 4.881658554077148,
+      "learning_rate": 2.025974025974026e-05,
+      "loss": 0.5017,
+      "step": 75
+    },
+    {
+      "epoch": 1.0,
+      "eval_accuracy": 0.7941176470588235,
+      "eval_combined_score": 0.8201357466063348,
+      "eval_f1": 0.8461538461538461,
+      "eval_loss": 0.42805391550064087,
+      "eval_runtime": 1.3728,
+      "eval_samples_per_second": 297.193,
+      "eval_steps_per_second": 3.642,
+      "step": 77
+    },
+    {
+      "epoch": 1.2987012987012987,
+      "grad_norm": 11.543705940246582,
+      "learning_rate": 1.7012987012987013e-05,
+      "loss": 0.4403,
+      "step": 100
+    },
+    {
+      "epoch": 1.6233766233766234,
+      "grad_norm": 5.57627010345459,
+      "learning_rate": 1.3766233766233767e-05,
+      "loss": 0.395,
+      "step": 125
+    },
+    {
+      "epoch": 1.948051948051948,
+      "grad_norm": 10.026427268981934,
+      "learning_rate": 1.051948051948052e-05,
+      "loss": 0.3935,
+      "step": 150
+    },
+    {
+      "epoch": 2.0,
+      "eval_accuracy": 0.8553921568627451,
+      "eval_combined_score": 0.8747337804062378,
+      "eval_f1": 0.8940754039497307,
+      "eval_loss": 0.34582728147506714,
+      "eval_runtime": 1.4011,
+      "eval_samples_per_second": 291.21,
+      "eval_steps_per_second": 3.569,
+      "step": 154
+    },
+    {
+      "epoch": 2.2727272727272725,
+      "grad_norm": 9.870122909545898,
+      "learning_rate": 7.272727272727273e-06,
+      "loss": 0.2911,
+      "step": 175
+    },
+    {
+      "epoch": 2.5974025974025974,
+      "grad_norm": 5.679322719573975,
+      "learning_rate": 4.025974025974026e-06,
+      "loss": 0.3041,
+      "step": 200
+    },
+    {
+      "epoch": 2.9220779220779223,
+      "grad_norm": 11.193693161010742,
+      "learning_rate": 7.792207792207793e-07,
+      "loss": 0.2776,
+      "step": 225
+    },
+    {
+      "epoch": 3.0,
+      "eval_accuracy": 0.8382352941176471,
+      "eval_combined_score": 0.859977862112587,
+      "eval_f1": 0.8817204301075269,
+      "eval_loss": 0.3466010093688965,
+      "eval_runtime": 1.397,
+      "eval_samples_per_second": 292.055,
+      "eval_steps_per_second": 3.579,
+      "step": 231
+    },
+    {
+      "epoch": 3.0,
+      "step": 231,
+      "total_flos": 625721256303600.0,
+      "train_loss": 0.41218561940378956,
+      "train_runtime": 20.9554,
+      "train_samples_per_second": 525.114,
+      "train_steps_per_second": 11.023
+    }
+  ],
+  "logging_steps": 25,
+  "max_steps": 231,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 3,
+  "save_steps": 500,
+  "stateful_callbacks": {
+    "TrainerControl": {
+      "args": {
+        "should_epoch_stop": false,
+        "should_evaluate": false,
+        "should_log": false,
+        "should_save": false,
+        "should_training_stop": false
+      },
+      "attributes": {}
+    }
+  },
+  "total_flos": 625721256303600.0,
+  "train_batch_size": 48,
+  "trial_name": null,
+  "trial_params": null
+}

mbert_rtx6000_metrics/result_ablation_mbert/mrpc/mBERT/hf_sequence_classifier/seed_44/all_results.json

@@ -0,0 +1,17 @@
+{
+    "epoch": 3.0,
+    "eval_accuracy": 0.8382352941176471,
+    "eval_combined_score": 0.8580065359477125,
+    "eval_f1": 0.8777777777777778,
+    "eval_loss": 0.35651159286499023,
+    "eval_runtime": 2.0928,
+    "eval_samples": 408,
+    "eval_samples_per_second": 194.951,
+    "eval_steps_per_second": 2.389,
+    "total_flos": 626644200552480.0,
+    "train_loss": 0.4211170962362578,
+    "train_runtime": 29.053,
+    "train_samples": 3668,
+    "train_samples_per_second": 378.756,
+    "train_steps_per_second": 7.951
+}

mbert_rtx6000_metrics/result_ablation_mbert/mrpc/mBERT/hf_sequence_classifier/seed_44/eval_results.json

@@ -0,0 +1,11 @@
+{
+    "epoch": 3.0,
+    "eval_accuracy": 0.8382352941176471,
+    "eval_combined_score": 0.8580065359477125,
+    "eval_f1": 0.8777777777777778,
+    "eval_loss": 0.35651159286499023,
+    "eval_runtime": 2.0928,
+    "eval_samples": 408,
+    "eval_samples_per_second": 194.951,
+    "eval_steps_per_second": 2.389
+}

mbert_rtx6000_metrics/result_ablation_mbert/mrpc/mBERT/hf_sequence_classifier/seed_44/run_command.txt

@@ -0,0 +1 @@
+/usr/local/bin/python /workspace/run_glue_MHA_gated.py --model_name_or_path bert-base-multilingual-cased --dataset_name tmnam20/ViGLUE --task_name mrpc --max_seq_length 128 --do_train --do_eval --evaluation_strategy epoch --save_strategy no --logging_strategy steps --logging_steps 25 --report_to none --num_train_epochs 3 --learning_rate 3e-05 --per_device_train_batch_size 48 --per_device_eval_batch_size 96 --gradient_accumulation_steps 1 --dataloader_num_workers 1 --seed 44 --output_dir result_ablation_mbert/mrpc/mBERT/hf_sequence_classifier/seed_44 --overwrite_output_dir --trust_remote_code true --bf16 --tf32 true --skip_model_save

mbert_rtx6000_metrics/result_ablation_mbert/mrpc/mBERT/hf_sequence_classifier/seed_44/train_results.json

@@ -0,0 +1,9 @@
+{
+    "epoch": 3.0,
+    "total_flos": 626644200552480.0,
+    "train_loss": 0.4211170962362578,
+    "train_runtime": 29.053,
+    "train_samples": 3668,
+    "train_samples_per_second": 378.756,
+    "train_steps_per_second": 7.951
+}

mbert_rtx6000_metrics/result_ablation_mbert/mrpc/mBERT/hf_sequence_classifier/seed_44/trainer_state.json

@@ -0,0 +1,138 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 3.0,
+  "eval_steps": 500,
+  "global_step": 231,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.3246753246753247,
+      "grad_norm": 2.104104995727539,
+      "learning_rate": 2.6753246753246755e-05,
+      "loss": 0.6058,
+      "step": 25
+    },
+    {
+      "epoch": 0.6493506493506493,
+      "grad_norm": 3.205981969833374,
+      "learning_rate": 2.350649350649351e-05,
+      "loss": 0.5493,
+      "step": 50
+    },
+    {
+      "epoch": 0.974025974025974,
+      "grad_norm": 6.8437323570251465,
+      "learning_rate": 2.025974025974026e-05,
+      "loss": 0.5066,
+      "step": 75
+    },
+    {
+      "epoch": 1.0,
+      "eval_accuracy": 0.7818627450980392,
+      "eval_combined_score": 0.8031601693932011,
+      "eval_f1": 0.8244575936883629,
+      "eval_loss": 0.4698929190635681,
+      "eval_runtime": 1.4776,
+      "eval_samples_per_second": 276.116,
+      "eval_steps_per_second": 3.384,
+      "step": 77
+    },
+    {
+      "epoch": 1.2987012987012987,
+      "grad_norm": 6.096251487731934,
+      "learning_rate": 1.7012987012987013e-05,
+      "loss": 0.4229,
+      "step": 100
+    },
+    {
+      "epoch": 1.6233766233766234,
+      "grad_norm": 5.171522617340088,
+      "learning_rate": 1.3766233766233767e-05,
+      "loss": 0.409,
+      "step": 125
+    },
+    {
+      "epoch": 1.948051948051948,
+      "grad_norm": 4.848453521728516,
+      "learning_rate": 1.051948051948052e-05,
+      "loss": 0.4078,
+      "step": 150
+    },
+    {
+      "epoch": 2.0,
+      "eval_accuracy": 0.8235294117647058,
+      "eval_combined_score": 0.8453440416757109,
+      "eval_f1": 0.8671586715867159,
+      "eval_loss": 0.36354437470436096,
+      "eval_runtime": 1.8924,
+      "eval_samples_per_second": 215.603,
+      "eval_steps_per_second": 2.642,
+      "step": 154
+    },
+    {
+      "epoch": 2.2727272727272725,
+      "grad_norm": 10.848043441772461,
+      "learning_rate": 7.272727272727273e-06,
+      "loss": 0.3305,
+      "step": 175
+    },
+    {
+      "epoch": 2.5974025974025974,
+      "grad_norm": 6.4878692626953125,
+      "learning_rate": 4.025974025974026e-06,
+      "loss": 0.2938,
+      "step": 200
+    },
+    {
+      "epoch": 2.9220779220779223,
+      "grad_norm": 4.981741905212402,
+      "learning_rate": 7.792207792207793e-07,
+      "loss": 0.2868,
+      "step": 225
+    },
+    {
+      "epoch": 3.0,
+      "eval_accuracy": 0.8382352941176471,
+      "eval_combined_score": 0.8580065359477125,
+      "eval_f1": 0.8777777777777778,
+      "eval_loss": 0.35651159286499023,
+      "eval_runtime": 2.1444,
+      "eval_samples_per_second": 190.262,
+      "eval_steps_per_second": 2.332,
+      "step": 231
+    },
+    {
+      "epoch": 3.0,
+      "step": 231,
+      "total_flos": 626644200552480.0,
+      "train_loss": 0.4211170962362578,
+      "train_runtime": 29.053,
+      "train_samples_per_second": 378.756,
+      "train_steps_per_second": 7.951
+    }
+  ],
+  "logging_steps": 25,
+  "max_steps": 231,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 3,
+  "save_steps": 500,
+  "stateful_callbacks": {
+    "TrainerControl": {
+      "args": {
+        "should_epoch_stop": false,
+        "should_evaluate": false,
+        "should_log": false,
+        "should_save": false,
+        "should_training_stop": false
+      },
+      "attributes": {}
+    }
+  },
+  "total_flos": 626644200552480.0,
+  "train_batch_size": 48,
+  "trial_name": null,
+  "trial_params": null
+}