Upload batch 7/24: mrpc/mBERT/attention/seed_42 ... mrpc/mBERT/cls/seed_42

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

@@ -0,0 +1,17 @@
+{
+    "epoch": 3.0,
+    "eval_accuracy": 0.821078431372549,
+    "eval_combined_score": 0.8452440457399417,
+    "eval_f1": 0.8694096601073346,
+    "eval_loss": 0.39012983441352844,
+    "eval_runtime": 1.2955,
+    "eval_samples": 408,
+    "eval_samples_per_second": 314.927,
+    "eval_steps_per_second": 3.859,
+    "total_flos": 630624795082152.0,
+    "train_loss": 0.38344859509241014,
+    "train_runtime": 32.7775,
+    "train_samples": 3668,
+    "train_samples_per_second": 335.718,
+    "train_steps_per_second": 7.048
+}

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

@@ -0,0 +1,11 @@
+{
+    "epoch": 3.0,
+    "eval_accuracy": 0.821078431372549,
+    "eval_combined_score": 0.8452440457399417,
+    "eval_f1": 0.8694096601073346,
+    "eval_loss": 0.39012983441352844,
+    "eval_runtime": 1.2955,
+    "eval_samples": 408,
+    "eval_samples_per_second": 314.927,
+    "eval_steps_per_second": 3.859
+}

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

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

@@ -0,0 +1,9 @@
+{
+    "epoch": 3.0,
+    "total_flos": 630624795082152.0,
+    "train_loss": 0.38344859509241014,
+    "train_runtime": 32.7775,
+    "train_samples": 3668,
+    "train_samples_per_second": 335.718,
+    "train_steps_per_second": 7.048
+}

mbert_rtx6000_metrics/result_ablation_mbert/mrpc/mBERT/attention/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": 1.8919161558151245,
+      "learning_rate": 2.6753246753246755e-05,
+      "loss": 0.6274,
+      "step": 25
+    },
+    {
+      "epoch": 0.6493506493506493,
+      "grad_norm": 3.552095890045166,
+      "learning_rate": 2.350649350649351e-05,
+      "loss": 0.5334,
+      "step": 50
+    },
+    {
+      "epoch": 0.974025974025974,
+      "grad_norm": 13.546927452087402,
+      "learning_rate": 2.025974025974026e-05,
+      "loss": 0.4812,
+      "step": 75
+    },
+    {
+      "epoch": 1.0,
+      "eval_accuracy": 0.8186274509803921,
+      "eval_combined_score": 0.8446284108048814,
+      "eval_f1": 0.8706293706293706,
+      "eval_loss": 0.3727538287639618,
+      "eval_runtime": 1.8977,
+      "eval_samples_per_second": 215.002,
+      "eval_steps_per_second": 2.635,
+      "step": 77
+    },
+    {
+      "epoch": 1.2987012987012987,
+      "grad_norm": 6.742566108703613,
+      "learning_rate": 1.7012987012987013e-05,
+      "loss": 0.3793,
+      "step": 100
+    },
+    {
+      "epoch": 1.6233766233766234,
+      "grad_norm": 4.569272994995117,
+      "learning_rate": 1.3766233766233767e-05,
+      "loss": 0.3718,
+      "step": 125
+    },
+    {
+      "epoch": 1.948051948051948,
+      "grad_norm": 6.266359806060791,
+      "learning_rate": 1.051948051948052e-05,
+      "loss": 0.3448,
+      "step": 150
+    },
+    {
+      "epoch": 2.0,
+      "eval_accuracy": 0.8259803921568627,
+      "eval_combined_score": 0.8499351339114687,
+      "eval_f1": 0.8738898756660746,
+      "eval_loss": 0.35563987493515015,
+      "eval_runtime": 1.3779,
+      "eval_samples_per_second": 296.106,
+      "eval_steps_per_second": 3.629,
+      "step": 154
+    },
+    {
+      "epoch": 2.2727272727272725,
+      "grad_norm": 3.569977283477783,
+      "learning_rate": 7.272727272727273e-06,
+      "loss": 0.2726,
+      "step": 175
+    },
+    {
+      "epoch": 2.5974025974025974,
+      "grad_norm": 8.519720077514648,
+      "learning_rate": 4.025974025974026e-06,
+      "loss": 0.2367,
+      "step": 200
+    },
+    {
+      "epoch": 2.9220779220779223,
+      "grad_norm": 7.060443878173828,
+      "learning_rate": 7.792207792207793e-07,
+      "loss": 0.2367,
+      "step": 225
+    },
+    {
+      "epoch": 3.0,
+      "eval_accuracy": 0.821078431372549,
+      "eval_combined_score": 0.8452440457399417,
+      "eval_f1": 0.8694096601073346,
+      "eval_loss": 0.39012983441352844,
+      "eval_runtime": 1.2444,
+      "eval_samples_per_second": 327.87,
+      "eval_steps_per_second": 4.018,
+      "step": 231
+    },
+    {
+      "epoch": 3.0,
+      "step": 231,
+      "total_flos": 630624795082152.0,
+      "train_loss": 0.38344859509241014,
+      "train_runtime": 32.7775,
+      "train_samples_per_second": 335.718,
+      "train_steps_per_second": 7.048
+    }
+  ],
+  "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": 630624795082152.0,
+  "train_batch_size": 48,
+  "trial_name": null,
+  "trial_params": null
+}

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

@@ -0,0 +1,17 @@
+{
+    "epoch": 3.0,
+    "eval_accuracy": 0.8333333333333334,
+    "eval_combined_score": 0.8543956043956045,
+    "eval_f1": 0.8754578754578755,
+    "eval_loss": 0.33410194516181946,
+    "eval_runtime": 2.021,
+    "eval_samples": 408,
+    "eval_samples_per_second": 201.876,
+    "eval_steps_per_second": 2.474,
+    "total_flos": 628605196579320.0,
+    "train_loss": 0.39260981506083437,
+    "train_runtime": 28.6766,
+    "train_samples": 3668,
+    "train_samples_per_second": 383.728,
+    "train_steps_per_second": 8.055
+}

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

@@ -0,0 +1,11 @@
+{
+    "epoch": 3.0,
+    "eval_accuracy": 0.8333333333333334,
+    "eval_combined_score": 0.8543956043956045,
+    "eval_f1": 0.8754578754578755,
+    "eval_loss": 0.33410194516181946,
+    "eval_runtime": 2.021,
+    "eval_samples": 408,
+    "eval_samples_per_second": 201.876,
+    "eval_steps_per_second": 2.474
+}

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

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

@@ -0,0 +1,9 @@
+{
+    "epoch": 3.0,
+    "total_flos": 628605196579320.0,
+    "train_loss": 0.39260981506083437,
+    "train_runtime": 28.6766,
+    "train_samples": 3668,
+    "train_samples_per_second": 383.728,
+    "train_steps_per_second": 8.055
+}

mbert_rtx6000_metrics/result_ablation_mbert/mrpc/mBERT/attention/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": 3.246776819229126,
+      "learning_rate": 2.6753246753246755e-05,
+      "loss": 0.6006,
+      "step": 25
+    },
+    {
+      "epoch": 0.6493506493506493,
+      "grad_norm": 4.267175197601318,
+      "learning_rate": 2.350649350649351e-05,
+      "loss": 0.5232,
+      "step": 50
+    },
+    {
+      "epoch": 0.974025974025974,
+      "grad_norm": 2.849297285079956,
+      "learning_rate": 2.025974025974026e-05,
+      "loss": 0.4914,
+      "step": 75
+    },
+    {
+      "epoch": 1.0,
+      "eval_accuracy": 0.7965686274509803,
+      "eval_combined_score": 0.8318843137254901,
+      "eval_f1": 0.8672,
+      "eval_loss": 0.407106876373291,
+      "eval_runtime": 1.3234,
+      "eval_samples_per_second": 308.285,
+      "eval_steps_per_second": 3.778,
+      "step": 77
+    },
+    {
+      "epoch": 1.2987012987012987,
+      "grad_norm": 11.595773696899414,
+      "learning_rate": 1.7012987012987013e-05,
+      "loss": 0.3953,
+      "step": 100
+    },
+    {
+      "epoch": 1.6233766233766234,
+      "grad_norm": 4.524789810180664,
+      "learning_rate": 1.3766233766233767e-05,
+      "loss": 0.4084,
+      "step": 125
+    },
+    {
+      "epoch": 1.948051948051948,
+      "grad_norm": 12.777288436889648,
+      "learning_rate": 1.051948051948052e-05,
+      "loss": 0.3733,
+      "step": 150
+    },
+    {
+      "epoch": 2.0,
+      "eval_accuracy": 0.8284313725490197,
+      "eval_combined_score": 0.8486726151134611,
+      "eval_f1": 0.8689138576779026,
+      "eval_loss": 0.35035502910614014,
+      "eval_runtime": 1.5875,
+      "eval_samples_per_second": 257.009,
+      "eval_steps_per_second": 3.15,
+      "step": 154
+    },
+    {
+      "epoch": 2.2727272727272725,
+      "grad_norm": 5.481242656707764,
+      "learning_rate": 7.272727272727273e-06,
+      "loss": 0.2687,
+      "step": 175
+    },
+    {
+      "epoch": 2.5974025974025974,
+      "grad_norm": 6.9631805419921875,
+      "learning_rate": 4.025974025974026e-06,
+      "loss": 0.2599,
+      "step": 200
+    },
+    {
+      "epoch": 2.9220779220779223,
+      "grad_norm": 13.768798828125,
+      "learning_rate": 7.792207792207793e-07,
+      "loss": 0.2515,
+      "step": 225
+    },
+    {
+      "epoch": 3.0,
+      "eval_accuracy": 0.8333333333333334,
+      "eval_combined_score": 0.8543956043956045,
+      "eval_f1": 0.8754578754578755,
+      "eval_loss": 0.33410194516181946,
+      "eval_runtime": 1.9704,
+      "eval_samples_per_second": 207.065,
+      "eval_steps_per_second": 2.538,
+      "step": 231
+    },
+    {
+      "epoch": 3.0,
+      "step": 231,
+      "total_flos": 628605196579320.0,
+      "train_loss": 0.39260981506083437,
+      "train_runtime": 28.6766,
+      "train_samples_per_second": 383.728,
+      "train_steps_per_second": 8.055
+    }
+  ],
+  "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": 628605196579320.0,
+  "train_batch_size": 48,
+  "trial_name": null,
+  "trial_params": null
+}

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

@@ -0,0 +1,17 @@
+{
+    "epoch": 3.0,
+    "eval_accuracy": 0.8578431372549019,
+    "eval_combined_score": 0.8773201451399064,
+    "eval_f1": 0.896797153024911,
+    "eval_loss": 0.3298549950122833,
+    "eval_runtime": 2.1403,
+    "eval_samples": 408,
+    "eval_samples_per_second": 190.628,
+    "eval_steps_per_second": 2.336,
+    "total_flos": 629532394664976.0,
+    "train_loss": 0.373855785890059,
+    "train_runtime": 36.2759,
+    "train_samples": 3668,
+    "train_samples_per_second": 303.342,
+    "train_steps_per_second": 6.368
+}

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

@@ -0,0 +1,11 @@
+{
+    "epoch": 3.0,
+    "eval_accuracy": 0.8578431372549019,
+    "eval_combined_score": 0.8773201451399064,
+    "eval_f1": 0.896797153024911,
+    "eval_loss": 0.3298549950122833,
+    "eval_runtime": 2.1403,
+    "eval_samples": 408,
+    "eval_samples_per_second": 190.628,
+    "eval_steps_per_second": 2.336
+}

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

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

@@ -0,0 +1,9 @@
+{
+    "epoch": 3.0,
+    "total_flos": 629532394664976.0,
+    "train_loss": 0.373855785890059,
+    "train_runtime": 36.2759,
+    "train_samples": 3668,
+    "train_samples_per_second": 303.342,
+    "train_steps_per_second": 6.368
+}

mbert_rtx6000_metrics/result_ablation_mbert/mrpc/mBERT/attention/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": 2.1300196647644043,
+      "learning_rate": 2.6753246753246755e-05,
+      "loss": 0.5958,
+      "step": 25
+    },
+    {
+      "epoch": 0.6493506493506493,
+      "grad_norm": 3.123434543609619,
+      "learning_rate": 2.350649350649351e-05,
+      "loss": 0.5298,
+      "step": 50
+    },
+    {
+      "epoch": 0.974025974025974,
+      "grad_norm": 5.821596145629883,
+      "learning_rate": 2.025974025974026e-05,
+      "loss": 0.4577,
+      "step": 75
+    },
+    {
+      "epoch": 1.0,
+      "eval_accuracy": 0.803921568627451,
+      "eval_combined_score": 0.8216394991731633,
+      "eval_f1": 0.8393574297188755,
+      "eval_loss": 0.40502360463142395,
+      "eval_runtime": 1.9812,
+      "eval_samples_per_second": 205.934,
+      "eval_steps_per_second": 2.524,
+      "step": 77
+    },
+    {
+      "epoch": 1.2987012987012987,
+      "grad_norm": 5.587197303771973,
+      "learning_rate": 1.7012987012987013e-05,
+      "loss": 0.3732,
+      "step": 100
+    },
+    {
+      "epoch": 1.6233766233766234,
+      "grad_norm": 3.907088279724121,
+      "learning_rate": 1.3766233766233767e-05,
+      "loss": 0.3712,
+      "step": 125
+    },
+    {
+      "epoch": 1.948051948051948,
+      "grad_norm": 4.280189037322998,
+      "learning_rate": 1.051948051948052e-05,
+      "loss": 0.3615,
+      "step": 150
+    },
+    {
+      "epoch": 2.0,
+      "eval_accuracy": 0.8602941176470589,
+      "eval_combined_score": 0.8795253891965312,
+      "eval_f1": 0.8987566607460036,
+      "eval_loss": 0.31799736618995667,
+      "eval_runtime": 1.881,
+      "eval_samples_per_second": 216.9,
+      "eval_steps_per_second": 2.658,
+      "step": 154
+    },
+    {
+      "epoch": 2.2727272727272725,
+      "grad_norm": 6.380615234375,
+      "learning_rate": 7.272727272727273e-06,
+      "loss": 0.2526,
+      "step": 175
+    },
+    {
+      "epoch": 2.5974025974025974,
+      "grad_norm": 6.170959949493408,
+      "learning_rate": 4.025974025974026e-06,
+      "loss": 0.2355,
+      "step": 200
+    },
+    {
+      "epoch": 2.9220779220779223,
+      "grad_norm": 3.874918222427368,
+      "learning_rate": 7.792207792207793e-07,
+      "loss": 0.2247,
+      "step": 225
+    },
+    {
+      "epoch": 3.0,
+      "eval_accuracy": 0.8578431372549019,
+      "eval_combined_score": 0.8773201451399064,
+      "eval_f1": 0.896797153024911,
+      "eval_loss": 0.3298549950122833,
+      "eval_runtime": 2.1261,
+      "eval_samples_per_second": 191.899,
+      "eval_steps_per_second": 2.352,
+      "step": 231
+    },
+    {
+      "epoch": 3.0,
+      "step": 231,
+      "total_flos": 629532394664976.0,
+      "train_loss": 0.373855785890059,
+      "train_runtime": 36.2759,
+      "train_samples_per_second": 303.342,
+      "train_steps_per_second": 6.368
+    }
+  ],
+  "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": 629532394664976.0,
+  "train_batch_size": 48,
+  "trial_name": null,
+  "trial_params": null
+}

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

@@ -0,0 +1,17 @@
+{
+    "epoch": 3.0,
+    "eval_accuracy": 0.8284313725490197,
+    "eval_combined_score": 0.8501131221719458,
+    "eval_f1": 0.8717948717948718,
+    "eval_loss": 0.3741132616996765,
+    "eval_runtime": 1.9516,
+    "eval_samples": 408,
+    "eval_samples_per_second": 209.057,
+    "eval_steps_per_second": 2.562,
+    "total_flos": 630619158933504.0,
+    "train_loss": 0.3877318664030595,
+    "train_runtime": 34.8202,
+    "train_samples": 3668,
+    "train_samples_per_second": 316.024,
+    "train_steps_per_second": 6.634
+}

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

@@ -0,0 +1,11 @@
+{
+    "epoch": 3.0,
+    "eval_accuracy": 0.8284313725490197,
+    "eval_combined_score": 0.8501131221719458,
+    "eval_f1": 0.8717948717948718,
+    "eval_loss": 0.3741132616996765,
+    "eval_runtime": 1.9516,
+    "eval_samples": 408,
+    "eval_samples_per_second": 209.057,
+    "eval_steps_per_second": 2.562
+}

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

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

@@ -0,0 +1,9 @@
+{
+    "epoch": 3.0,
+    "total_flos": 630619158933504.0,
+    "train_loss": 0.3877318664030595,
+    "train_runtime": 34.8202,
+    "train_samples": 3668,
+    "train_samples_per_second": 316.024,
+    "train_steps_per_second": 6.634
+}

mbert_rtx6000_metrics/result_ablation_mbert/mrpc/mBERT/cls/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": 5.96425199508667,
+      "learning_rate": 2.6753246753246755e-05,
+      "loss": 0.6056,
+      "step": 25
+    },
+    {
+      "epoch": 0.6493506493506493,
+      "grad_norm": 6.376526832580566,
+      "learning_rate": 2.350649350649351e-05,
+      "loss": 0.5115,
+      "step": 50
+    },
+    {
+      "epoch": 0.974025974025974,
+      "grad_norm": 3.437178611755371,
+      "learning_rate": 2.025974025974026e-05,
+      "loss": 0.5137,
+      "step": 75
+    },
+    {
+      "epoch": 1.0,
+      "eval_accuracy": 0.8186274509803921,
+      "eval_combined_score": 0.8378851540616246,
+      "eval_f1": 0.8571428571428571,
+      "eval_loss": 0.4310764968395233,
+      "eval_runtime": 2.016,
+      "eval_samples_per_second": 202.377,
+      "eval_steps_per_second": 2.48,
+      "step": 77
+    },
+    {
+      "epoch": 1.2987012987012987,
+      "grad_norm": 7.388753414154053,
+      "learning_rate": 1.7012987012987013e-05,
+      "loss": 0.3975,
+      "step": 100
+    },
+    {
+      "epoch": 1.6233766233766234,
+      "grad_norm": 4.554161548614502,
+      "learning_rate": 1.3766233766233767e-05,
+      "loss": 0.3889,
+      "step": 125
+    },
+    {
+      "epoch": 1.948051948051948,
+      "grad_norm": 6.1189751625061035,
+      "learning_rate": 1.051948051948052e-05,
+      "loss": 0.347,
+      "step": 150
+    },
+    {
+      "epoch": 2.0,
+      "eval_accuracy": 0.821078431372549,
+      "eval_combined_score": 0.8433200628317625,
+      "eval_f1": 0.8655616942909761,
+      "eval_loss": 0.3410874903202057,
+      "eval_runtime": 1.8925,
+      "eval_samples_per_second": 215.583,
+      "eval_steps_per_second": 2.642,
+      "step": 154
+    },
+    {
+      "epoch": 2.2727272727272725,
+      "grad_norm": 4.478530406951904,
+      "learning_rate": 7.272727272727273e-06,
+      "loss": 0.2785,
+      "step": 175
+    },
+    {
+      "epoch": 2.5974025974025974,
+      "grad_norm": 9.207025527954102,
+      "learning_rate": 4.025974025974026e-06,
+      "loss": 0.2466,
+      "step": 200
+    },
+    {
+      "epoch": 2.9220779220779223,
+      "grad_norm": 11.181907653808594,
+      "learning_rate": 7.792207792207793e-07,
+      "loss": 0.2329,
+      "step": 225
+    },
+    {
+      "epoch": 3.0,
+      "eval_accuracy": 0.8284313725490197,
+      "eval_combined_score": 0.8501131221719458,
+      "eval_f1": 0.8717948717948718,
+      "eval_loss": 0.3741132616996765,
+      "eval_runtime": 1.8733,
+      "eval_samples_per_second": 217.803,
+      "eval_steps_per_second": 2.669,
+      "step": 231
+    },
+    {
+      "epoch": 3.0,
+      "step": 231,
+      "total_flos": 630619158933504.0,
+      "train_loss": 0.3877318664030595,
+      "train_runtime": 34.8202,
+      "train_samples_per_second": 316.024,
+      "train_steps_per_second": 6.634
+    }
+  ],
+  "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
+}