Upload batch 10/24: mrpc/mBERT/max/seed_42 ... mrpc/mBERT/mean/seed_42

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

@@ -0,0 +1,17 @@
+{
+    "epoch": 3.0,
+    "eval_accuracy": 0.8357843137254902,
+    "eval_combined_score": 0.8570936087683713,
+    "eval_f1": 0.8784029038112523,
+    "eval_loss": 0.3859576880931854,
+    "eval_runtime": 1.8922,
+    "eval_samples": 408,
+    "eval_samples_per_second": 215.619,
+    "eval_steps_per_second": 2.642,
+    "total_flos": 630619158933504.0,
+    "train_loss": 0.37803170020446114,
+    "train_runtime": 35.9479,
+    "train_samples": 3668,
+    "train_samples_per_second": 306.11,
+    "train_steps_per_second": 6.426
+}

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

@@ -0,0 +1,11 @@
+{
+    "epoch": 3.0,
+    "eval_accuracy": 0.8357843137254902,
+    "eval_combined_score": 0.8570936087683713,
+    "eval_f1": 0.8784029038112523,
+    "eval_loss": 0.3859576880931854,
+    "eval_runtime": 1.8922,
+    "eval_samples": 408,
+    "eval_samples_per_second": 215.619,
+    "eval_steps_per_second": 2.642
+}

mbert_rtx6000_metrics/result_ablation_mbert/mrpc/mBERT/max/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/max/seed_42 --overwrite_output_dir --trust_remote_code true --bf16 --tf32 true --skip_model_save --use_mha true --pooling_strategy max --n_heads 4 --head_dim 32 --classifier_hidden_dim 512 --head_dropout 0.05

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

@@ -0,0 +1,9 @@
+{
+    "epoch": 3.0,
+    "total_flos": 630619158933504.0,
+    "train_loss": 0.37803170020446114,
+    "train_runtime": 35.9479,
+    "train_samples": 3668,
+    "train_samples_per_second": 306.11,
+    "train_steps_per_second": 6.426
+}

mbert_rtx6000_metrics/result_ablation_mbert/mrpc/mBERT/max/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": 2.598341226577759,
+      "learning_rate": 2.6753246753246755e-05,
+      "loss": 0.5985,
+      "step": 25
+    },
+    {
+      "epoch": 0.6493506493506493,
+      "grad_norm": 7.255716323852539,
+      "learning_rate": 2.350649350649351e-05,
+      "loss": 0.479,
+      "step": 50
+    },
+    {
+      "epoch": 0.974025974025974,
+      "grad_norm": 3.670036792755127,
+      "learning_rate": 2.025974025974026e-05,
+      "loss": 0.4684,
+      "step": 75
+    },
+    {
+      "epoch": 1.0,
+      "eval_accuracy": 0.803921568627451,
+      "eval_combined_score": 0.8253324318233041,
+      "eval_f1": 0.8467432950191571,
+      "eval_loss": 0.39329349994659424,
+      "eval_runtime": 2.0906,
+      "eval_samples_per_second": 195.162,
+      "eval_steps_per_second": 2.392,
+      "step": 77
+    },
+    {
+      "epoch": 1.2987012987012987,
+      "grad_norm": 5.792549133300781,
+      "learning_rate": 1.7012987012987013e-05,
+      "loss": 0.3743,
+      "step": 100
+    },
+    {
+      "epoch": 1.6233766233766234,
+      "grad_norm": 9.58816909790039,
+      "learning_rate": 1.3766233766233767e-05,
+      "loss": 0.3846,
+      "step": 125
+    },
+    {
+      "epoch": 1.948051948051948,
+      "grad_norm": 5.964817523956299,
+      "learning_rate": 1.051948051948052e-05,
+      "loss": 0.3478,
+      "step": 150
+    },
+    {
+      "epoch": 2.0,
+      "eval_accuracy": 0.8284313725490197,
+      "eval_combined_score": 0.849400871459695,
+      "eval_f1": 0.8703703703703703,
+      "eval_loss": 0.350210964679718,
+      "eval_runtime": 2.1974,
+      "eval_samples_per_second": 185.675,
+      "eval_steps_per_second": 2.275,
+      "step": 154
+    },
+    {
+      "epoch": 2.2727272727272725,
+      "grad_norm": 4.463625431060791,
+      "learning_rate": 7.272727272727273e-06,
+      "loss": 0.2867,
+      "step": 175
+    },
+    {
+      "epoch": 2.5974025974025974,
+      "grad_norm": 8.163609504699707,
+      "learning_rate": 4.025974025974026e-06,
+      "loss": 0.2565,
+      "step": 200
+    },
+    {
+      "epoch": 2.9220779220779223,
+      "grad_norm": 6.111831188201904,
+      "learning_rate": 7.792207792207793e-07,
+      "loss": 0.2384,
+      "step": 225
+    },
+    {
+      "epoch": 3.0,
+      "eval_accuracy": 0.8357843137254902,
+      "eval_combined_score": 0.8570936087683713,
+      "eval_f1": 0.8784029038112523,
+      "eval_loss": 0.3859576880931854,
+      "eval_runtime": 1.972,
+      "eval_samples_per_second": 206.893,
+      "eval_steps_per_second": 2.535,
+      "step": 231
+    },
+    {
+      "epoch": 3.0,
+      "step": 231,
+      "total_flos": 630619158933504.0,
+      "train_loss": 0.37803170020446114,
+      "train_runtime": 35.9479,
+      "train_samples_per_second": 306.11,
+      "train_steps_per_second": 6.426
+    }
+  ],
+  "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": 630619158933504.0,
+  "train_batch_size": 48,
+  "trial_name": null,
+  "trial_params": null
+}

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

@@ -0,0 +1,17 @@
+{
+    "epoch": 3.0,
+    "eval_accuracy": 0.8529411764705882,
+    "eval_combined_score": 0.8717260626878489,
+    "eval_f1": 0.8905109489051095,
+    "eval_loss": 0.3556552827358246,
+    "eval_runtime": 1.6058,
+    "eval_samples": 408,
+    "eval_samples_per_second": 254.085,
+    "eval_steps_per_second": 3.114,
+    "total_flos": 628599578480640.0,
+    "train_loss": 0.3737406565513446,
+    "train_runtime": 38.0938,
+    "train_samples": 3668,
+    "train_samples_per_second": 288.866,
+    "train_steps_per_second": 6.064
+}

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

@@ -0,0 +1,11 @@
+{
+    "epoch": 3.0,
+    "eval_accuracy": 0.8529411764705882,
+    "eval_combined_score": 0.8717260626878489,
+    "eval_f1": 0.8905109489051095,
+    "eval_loss": 0.3556552827358246,
+    "eval_runtime": 1.6058,
+    "eval_samples": 408,
+    "eval_samples_per_second": 254.085,
+    "eval_steps_per_second": 3.114
+}

mbert_rtx6000_metrics/result_ablation_mbert/mrpc/mBERT/max/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/max/seed_43 --overwrite_output_dir --trust_remote_code true --bf16 --tf32 true --skip_model_save --use_mha true --pooling_strategy max --n_heads 4 --head_dim 32 --classifier_hidden_dim 512 --head_dropout 0.05

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

@@ -0,0 +1,9 @@
+{
+    "epoch": 3.0,
+    "total_flos": 628599578480640.0,
+    "train_loss": 0.3737406565513446,
+    "train_runtime": 38.0938,
+    "train_samples": 3668,
+    "train_samples_per_second": 288.866,
+    "train_steps_per_second": 6.064
+}

mbert_rtx6000_metrics/result_ablation_mbert/mrpc/mBERT/max/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": 2.613507032394409,
+      "learning_rate": 2.6753246753246755e-05,
+      "loss": 0.5876,
+      "step": 25
+    },
+    {
+      "epoch": 0.6493506493506493,
+      "grad_norm": 2.779635429382324,
+      "learning_rate": 2.350649350649351e-05,
+      "loss": 0.5059,
+      "step": 50
+    },
+    {
+      "epoch": 0.974025974025974,
+      "grad_norm": 4.840301036834717,
+      "learning_rate": 2.025974025974026e-05,
+      "loss": 0.4632,
+      "step": 75
+    },
+    {
+      "epoch": 1.0,
+      "eval_accuracy": 0.8333333333333334,
+      "eval_combined_score": 0.8552948255114321,
+      "eval_f1": 0.8772563176895307,
+      "eval_loss": 0.360188364982605,
+      "eval_runtime": 1.8153,
+      "eval_samples_per_second": 224.757,
+      "eval_steps_per_second": 2.754,
+      "step": 77
+    },
+    {
+      "epoch": 1.2987012987012987,
+      "grad_norm": 5.261049747467041,
+      "learning_rate": 1.7012987012987013e-05,
+      "loss": 0.3803,
+      "step": 100
+    },
+    {
+      "epoch": 1.6233766233766234,
+      "grad_norm": 5.005692481994629,
+      "learning_rate": 1.3766233766233767e-05,
+      "loss": 0.3665,
+      "step": 125
+    },
+    {
+      "epoch": 1.948051948051948,
+      "grad_norm": 13.645220756530762,
+      "learning_rate": 1.051948051948052e-05,
+      "loss": 0.3366,
+      "step": 150
+    },
+    {
+      "epoch": 2.0,
+      "eval_accuracy": 0.8259803921568627,
+      "eval_combined_score": 0.8461352054946272,
+      "eval_f1": 0.8662900188323918,
+      "eval_loss": 0.3570290505886078,
+      "eval_runtime": 2.0203,
+      "eval_samples_per_second": 201.946,
+      "eval_steps_per_second": 2.475,
+      "step": 154
+    },
+    {
+      "epoch": 2.2727272727272725,
+      "grad_norm": 5.498772621154785,
+      "learning_rate": 7.272727272727273e-06,
+      "loss": 0.2505,
+      "step": 175
+    },
+    {
+      "epoch": 2.5974025974025974,
+      "grad_norm": 12.274273872375488,
+      "learning_rate": 4.025974025974026e-06,
+      "loss": 0.2541,
+      "step": 200
+    },
+    {
+      "epoch": 2.9220779220779223,
+      "grad_norm": 14.405874252319336,
+      "learning_rate": 7.792207792207793e-07,
+      "loss": 0.2505,
+      "step": 225
+    },
+    {
+      "epoch": 3.0,
+      "eval_accuracy": 0.8529411764705882,
+      "eval_combined_score": 0.8717260626878489,
+      "eval_f1": 0.8905109489051095,
+      "eval_loss": 0.3556552827358246,
+      "eval_runtime": 1.6412,
+      "eval_samples_per_second": 248.594,
+      "eval_steps_per_second": 3.046,
+      "step": 231
+    },
+    {
+      "epoch": 3.0,
+      "step": 231,
+      "total_flos": 628599578480640.0,
+      "train_loss": 0.3737406565513446,
+      "train_runtime": 38.0938,
+      "train_samples_per_second": 288.866,
+      "train_steps_per_second": 6.064
+    }
+  ],
+  "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": 628599578480640.0,
+  "train_batch_size": 48,
+  "trial_name": null,
+  "trial_params": null
+}

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

@@ -0,0 +1,17 @@
+{
+    "epoch": 3.0,
+    "eval_accuracy": 0.8627450980392157,
+    "eval_combined_score": 0.8811933375500738,
+    "eval_f1": 0.899641577060932,
+    "eval_loss": 0.3449813723564148,
+    "eval_runtime": 1.4962,
+    "eval_samples": 408,
+    "eval_samples_per_second": 272.692,
+    "eval_steps_per_second": 3.342,
+    "total_flos": 629526768279552.0,
+    "train_loss": 0.3843384521864193,
+    "train_runtime": 35.4031,
+    "train_samples": 3668,
+    "train_samples_per_second": 310.82,
+    "train_steps_per_second": 6.525
+}

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

@@ -0,0 +1,11 @@
+{
+    "epoch": 3.0,
+    "eval_accuracy": 0.8627450980392157,
+    "eval_combined_score": 0.8811933375500738,
+    "eval_f1": 0.899641577060932,
+    "eval_loss": 0.3449813723564148,
+    "eval_runtime": 1.4962,
+    "eval_samples": 408,
+    "eval_samples_per_second": 272.692,
+    "eval_steps_per_second": 3.342
+}

mbert_rtx6000_metrics/result_ablation_mbert/mrpc/mBERT/max/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/max/seed_44 --overwrite_output_dir --trust_remote_code true --bf16 --tf32 true --skip_model_save --use_mha true --pooling_strategy max --n_heads 4 --head_dim 32 --classifier_hidden_dim 512 --head_dropout 0.05

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

@@ -0,0 +1,9 @@
+{
+    "epoch": 3.0,
+    "total_flos": 629526768279552.0,
+    "train_loss": 0.3843384521864193,
+    "train_runtime": 35.4031,
+    "train_samples": 3668,
+    "train_samples_per_second": 310.82,
+    "train_steps_per_second": 6.525
+}

mbert_rtx6000_metrics/result_ablation_mbert/mrpc/mBERT/max/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": 1.7536025047302246,
+      "learning_rate": 2.6753246753246755e-05,
+      "loss": 0.6077,
+      "step": 25
+    },
+    {
+      "epoch": 0.6493506493506493,
+      "grad_norm": 3.8616058826446533,
+      "learning_rate": 2.350649350649351e-05,
+      "loss": 0.5398,
+      "step": 50
+    },
+    {
+      "epoch": 0.974025974025974,
+      "grad_norm": 4.83195686340332,
+      "learning_rate": 2.025974025974026e-05,
+      "loss": 0.4447,
+      "step": 75
+    },
+    {
+      "epoch": 1.0,
+      "eval_accuracy": 0.7573529411764706,
+      "eval_combined_score": 0.7749028856825749,
+      "eval_f1": 0.7924528301886793,
+      "eval_loss": 0.4518143832683563,
+      "eval_runtime": 2.0675,
+      "eval_samples_per_second": 197.337,
+      "eval_steps_per_second": 2.418,
+      "step": 77
+    },
+    {
+      "epoch": 1.2987012987012987,
+      "grad_norm": 7.229933261871338,
+      "learning_rate": 1.7012987012987013e-05,
+      "loss": 0.385,
+      "step": 100
+    },
+    {
+      "epoch": 1.6233766233766234,
+      "grad_norm": 5.55099630355835,
+      "learning_rate": 1.3766233766233767e-05,
+      "loss": 0.3622,
+      "step": 125
+    },
+    {
+      "epoch": 1.948051948051948,
+      "grad_norm": 5.723285675048828,
+      "learning_rate": 1.051948051948052e-05,
+      "loss": 0.3733,
+      "step": 150
+    },
+    {
+      "epoch": 2.0,
+      "eval_accuracy": 0.8455882352941176,
+      "eval_combined_score": 0.8658320923306031,
+      "eval_f1": 0.8860759493670886,
+      "eval_loss": 0.33651044964790344,
+      "eval_runtime": 1.5903,
+      "eval_samples_per_second": 256.549,
+      "eval_steps_per_second": 3.144,
+      "step": 154
+    },
+    {
+      "epoch": 2.2727272727272725,
+      "grad_norm": 7.109714031219482,
+      "learning_rate": 7.272727272727273e-06,
+      "loss": 0.2741,
+      "step": 175
+    },
+    {
+      "epoch": 2.5974025974025974,
+      "grad_norm": 7.071986675262451,
+      "learning_rate": 4.025974025974026e-06,
+      "loss": 0.2498,
+      "step": 200
+    },
+    {
+      "epoch": 2.9220779220779223,
+      "grad_norm": 3.7733166217803955,
+      "learning_rate": 7.792207792207793e-07,
+      "loss": 0.258,
+      "step": 225
+    },
+    {
+      "epoch": 3.0,
+      "eval_accuracy": 0.8627450980392157,
+      "eval_combined_score": 0.8811933375500738,
+      "eval_f1": 0.899641577060932,
+      "eval_loss": 0.3449813723564148,
+      "eval_runtime": 1.3102,
+      "eval_samples_per_second": 311.407,
+      "eval_steps_per_second": 3.816,
+      "step": 231
+    },
+    {
+      "epoch": 3.0,
+      "step": 231,
+      "total_flos": 629526768279552.0,
+      "train_loss": 0.3843384521864193,
+      "train_runtime": 35.4031,
+      "train_samples_per_second": 310.82,
+      "train_steps_per_second": 6.525
+    }
+  ],
+  "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": 629526768279552.0,
+  "train_batch_size": 48,
+  "trial_name": null,
+  "trial_params": null
+}

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

@@ -0,0 +1,17 @@
+{
+    "epoch": 3.0,
+    "eval_accuracy": 0.8406862745098039,
+    "eval_combined_score": 0.8617845786963434,
+    "eval_f1": 0.8828828828828829,
+    "eval_loss": 0.3512035310268402,
+    "eval_runtime": 1.4064,
+    "eval_samples": 408,
+    "eval_samples_per_second": 290.1,
+    "eval_steps_per_second": 3.555,
+    "total_flos": 630619158933504.0,
+    "train_loss": 0.38461016886162036,
+    "train_runtime": 32.4466,
+    "train_samples": 3668,
+    "train_samples_per_second": 339.142,
+    "train_steps_per_second": 7.119
+}

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

@@ -0,0 +1,11 @@
+{
+    "epoch": 3.0,
+    "eval_accuracy": 0.8406862745098039,
+    "eval_combined_score": 0.8617845786963434,
+    "eval_f1": 0.8828828828828829,
+    "eval_loss": 0.3512035310268402,
+    "eval_runtime": 1.4064,
+    "eval_samples": 408,
+    "eval_samples_per_second": 290.1,
+    "eval_steps_per_second": 3.555
+}

mbert_rtx6000_metrics/result_ablation_mbert/mrpc/mBERT/mean/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/mean/seed_42 --overwrite_output_dir --trust_remote_code true --bf16 --tf32 true --skip_model_save --use_mha true --pooling_strategy mean --n_heads 4 --head_dim 32 --classifier_hidden_dim 512 --head_dropout 0.05

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

@@ -0,0 +1,9 @@
+{
+    "epoch": 3.0,
+    "total_flos": 630619158933504.0,
+    "train_loss": 0.38461016886162036,
+    "train_runtime": 32.4466,
+    "train_samples": 3668,
+    "train_samples_per_second": 339.142,
+    "train_steps_per_second": 7.119
+}

mbert_rtx6000_metrics/result_ablation_mbert/mrpc/mBERT/mean/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": 2.099273681640625,
+      "learning_rate": 2.6753246753246755e-05,
+      "loss": 0.6143,
+      "step": 25
+    },
+    {
+      "epoch": 0.6493506493506493,
+      "grad_norm": 8.029365539550781,
+      "learning_rate": 2.350649350649351e-05,
+      "loss": 0.5179,
+      "step": 50
+    },
+    {
+      "epoch": 0.974025974025974,
+      "grad_norm": 3.3279473781585693,
+      "learning_rate": 2.025974025974026e-05,
+      "loss": 0.5085,
+      "step": 75
+    },
+    {
+      "epoch": 1.0,
+      "eval_accuracy": 0.8259803921568627,
+      "eval_combined_score": 0.8458824456058416,
+      "eval_f1": 0.8657844990548205,
+      "eval_loss": 0.4006221890449524,
+      "eval_runtime": 1.6069,
+      "eval_samples_per_second": 253.904,
+      "eval_steps_per_second": 3.112,
+      "step": 77
+    },
+    {
+      "epoch": 1.2987012987012987,
+      "grad_norm": 6.741519927978516,
+      "learning_rate": 1.7012987012987013e-05,
+      "loss": 0.3848,
+      "step": 100
+    },
+    {
+      "epoch": 1.6233766233766234,
+      "grad_norm": 6.245471000671387,
+      "learning_rate": 1.3766233766233767e-05,
+      "loss": 0.388,
+      "step": 125
+    },
+    {
+      "epoch": 1.948051948051948,
+      "grad_norm": 7.738393306732178,
+      "learning_rate": 1.051948051948052e-05,
+      "loss": 0.3445,
+      "step": 150
+    },
+    {
+      "epoch": 2.0,
+      "eval_accuracy": 0.8553921568627451,
+      "eval_combined_score": 0.8752982098700937,
+      "eval_f1": 0.8952042628774423,
+      "eval_loss": 0.332045316696167,
+      "eval_runtime": 1.4556,
+      "eval_samples_per_second": 280.291,
+      "eval_steps_per_second": 3.435,
+      "step": 154
+    },
+    {
+      "epoch": 2.2727272727272725,
+      "grad_norm": 3.555630922317505,
+      "learning_rate": 7.272727272727273e-06,
+      "loss": 0.2704,
+      "step": 175
+    },
+    {
+      "epoch": 2.5974025974025974,
+      "grad_norm": 9.103394508361816,
+      "learning_rate": 4.025974025974026e-06,
+      "loss": 0.2436,
+      "step": 200
+    },
+    {
+      "epoch": 2.9220779220779223,
+      "grad_norm": 5.559789180755615,
+      "learning_rate": 7.792207792207793e-07,
+      "loss": 0.2216,
+      "step": 225
+    },
+    {
+      "epoch": 3.0,
+      "eval_accuracy": 0.8406862745098039,
+      "eval_combined_score": 0.8617845786963434,
+      "eval_f1": 0.8828828828828829,
+      "eval_loss": 0.3512035310268402,
+      "eval_runtime": 1.3617,
+      "eval_samples_per_second": 299.627,
+      "eval_steps_per_second": 3.672,
+      "step": 231
+    },
+    {
+      "epoch": 3.0,
+      "step": 231,
+      "total_flos": 630619158933504.0,
+      "train_loss": 0.38461016886162036,
+      "train_runtime": 32.4466,
+      "train_samples_per_second": 339.142,
+      "train_steps_per_second": 7.119
+    }
+  ],
+  "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": 630619158933504.0,
+  "train_batch_size": 48,
+  "trial_name": null,
+  "trial_params": null
+}