Upload batch 9/14: sst2/mBERT/gated_multi_branch/seed_44 ... sst2/mBERT/max/seed_44
Browse files- mbert_paper_metrics/result_ablation_mbert_paper/sst2/mBERT/gated_multi_branch/seed_44/all_results.json +15 -0
- mbert_paper_metrics/result_ablation_mbert_paper/sst2/mBERT/gated_multi_branch/seed_44/eval_results.json +9 -0
- mbert_paper_metrics/result_ablation_mbert_paper/sst2/mBERT/gated_multi_branch/seed_44/run_command.txt +1 -0
- mbert_paper_metrics/result_ablation_mbert_paper/sst2/mBERT/gated_multi_branch/seed_44/train_results.json +9 -0
- mbert_paper_metrics/result_ablation_mbert_paper/sst2/mBERT/gated_multi_branch/seed_44/trainer_state.json +419 -0
- mbert_paper_metrics/result_ablation_mbert_paper/sst2/mBERT/hf_sequence_classifier/seed_42/all_results.json +15 -0
- mbert_paper_metrics/result_ablation_mbert_paper/sst2/mBERT/hf_sequence_classifier/seed_42/eval_results.json +9 -0
- mbert_paper_metrics/result_ablation_mbert_paper/sst2/mBERT/hf_sequence_classifier/seed_42/run_command.txt +1 -0
- mbert_paper_metrics/result_ablation_mbert_paper/sst2/mBERT/hf_sequence_classifier/seed_42/train_results.json +9 -0
- mbert_paper_metrics/result_ablation_mbert_paper/sst2/mBERT/hf_sequence_classifier/seed_42/trainer_state.json +419 -0
- mbert_paper_metrics/result_ablation_mbert_paper/sst2/mBERT/hf_sequence_classifier/seed_43/all_results.json +15 -0
- mbert_paper_metrics/result_ablation_mbert_paper/sst2/mBERT/hf_sequence_classifier/seed_43/eval_results.json +9 -0
- mbert_paper_metrics/result_ablation_mbert_paper/sst2/mBERT/hf_sequence_classifier/seed_43/run_command.txt +1 -0
- mbert_paper_metrics/result_ablation_mbert_paper/sst2/mBERT/hf_sequence_classifier/seed_43/train_results.json +9 -0
- mbert_paper_metrics/result_ablation_mbert_paper/sst2/mBERT/hf_sequence_classifier/seed_43/trainer_state.json +419 -0
- mbert_paper_metrics/result_ablation_mbert_paper/sst2/mBERT/hf_sequence_classifier/seed_44/all_results.json +15 -0
- mbert_paper_metrics/result_ablation_mbert_paper/sst2/mBERT/hf_sequence_classifier/seed_44/eval_results.json +9 -0
- mbert_paper_metrics/result_ablation_mbert_paper/sst2/mBERT/hf_sequence_classifier/seed_44/run_command.txt +1 -0
- mbert_paper_metrics/result_ablation_mbert_paper/sst2/mBERT/hf_sequence_classifier/seed_44/train_results.json +9 -0
- mbert_paper_metrics/result_ablation_mbert_paper/sst2/mBERT/hf_sequence_classifier/seed_44/trainer_state.json +419 -0
- mbert_paper_metrics/result_ablation_mbert_paper/sst2/mBERT/max/seed_42/all_results.json +15 -0
- mbert_paper_metrics/result_ablation_mbert_paper/sst2/mBERT/max/seed_42/eval_results.json +9 -0
- mbert_paper_metrics/result_ablation_mbert_paper/sst2/mBERT/max/seed_42/run_command.txt +1 -0
- mbert_paper_metrics/result_ablation_mbert_paper/sst2/mBERT/max/seed_42/train_results.json +9 -0
- mbert_paper_metrics/result_ablation_mbert_paper/sst2/mBERT/max/seed_42/trainer_state.json +419 -0
- mbert_paper_metrics/result_ablation_mbert_paper/sst2/mBERT/max/seed_43/all_results.json +15 -0
- mbert_paper_metrics/result_ablation_mbert_paper/sst2/mBERT/max/seed_43/eval_results.json +9 -0
- mbert_paper_metrics/result_ablation_mbert_paper/sst2/mBERT/max/seed_43/run_command.txt +1 -0
- mbert_paper_metrics/result_ablation_mbert_paper/sst2/mBERT/max/seed_43/train_results.json +9 -0
- mbert_paper_metrics/result_ablation_mbert_paper/sst2/mBERT/max/seed_43/trainer_state.json +419 -0
- mbert_paper_metrics/result_ablation_mbert_paper/sst2/mBERT/max/seed_44/all_results.json +15 -0
- mbert_paper_metrics/result_ablation_mbert_paper/sst2/mBERT/max/seed_44/eval_results.json +9 -0
- mbert_paper_metrics/result_ablation_mbert_paper/sst2/mBERT/max/seed_44/run_command.txt +1 -0
- mbert_paper_metrics/result_ablation_mbert_paper/sst2/mBERT/max/seed_44/train_results.json +9 -0
- mbert_paper_metrics/result_ablation_mbert_paper/sst2/mBERT/max/seed_44/trainer_state.json +419 -0
mbert_paper_metrics/result_ablation_mbert_paper/sst2/mBERT/gated_multi_branch/seed_44/all_results.json
ADDED
|
@@ -0,0 +1,15 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 3.0,
|
| 3 |
+
"eval_accuracy": 0.875,
|
| 4 |
+
"eval_loss": 0.3715299665927887,
|
| 5 |
+
"eval_runtime": 2.1236,
|
| 6 |
+
"eval_samples": 872,
|
| 7 |
+
"eval_samples_per_second": 410.619,
|
| 8 |
+
"eval_steps_per_second": 2.825,
|
| 9 |
+
"total_flos": 6937506539055684.0,
|
| 10 |
+
"train_loss": 0.2534543433457448,
|
| 11 |
+
"train_runtime": 276.4842,
|
| 12 |
+
"train_samples": 67349,
|
| 13 |
+
"train_samples_per_second": 730.772,
|
| 14 |
+
"train_steps_per_second": 9.136
|
| 15 |
+
}
|
mbert_paper_metrics/result_ablation_mbert_paper/sst2/mBERT/gated_multi_branch/seed_44/eval_results.json
ADDED
|
@@ -0,0 +1,9 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 3.0,
|
| 3 |
+
"eval_accuracy": 0.875,
|
| 4 |
+
"eval_loss": 0.3715299665927887,
|
| 5 |
+
"eval_runtime": 2.1236,
|
| 6 |
+
"eval_samples": 872,
|
| 7 |
+
"eval_samples_per_second": 410.619,
|
| 8 |
+
"eval_steps_per_second": 2.825
|
| 9 |
+
}
|
mbert_paper_metrics/result_ablation_mbert_paper/sst2/mBERT/gated_multi_branch/seed_44/run_command.txt
ADDED
|
@@ -0,0 +1 @@
|
|
|
|
|
|
|
| 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 sst2 --max_seq_length 256 --do_train --do_eval --evaluation_strategy epoch --save_strategy no --logging_strategy steps --logging_steps 50 --report_to none --num_train_epochs 3 --learning_rate 2e-05 --per_device_train_batch_size 80 --per_device_eval_batch_size 168 --gradient_accumulation_steps 1 --dataloader_num_workers 2 --seed 44 --output_dir result_ablation_mbert_paper/sst2/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 8 --head_dim 32 --classifier_hidden_dim 512 --head_dropout 0.05
|
mbert_paper_metrics/result_ablation_mbert_paper/sst2/mBERT/gated_multi_branch/seed_44/train_results.json
ADDED
|
@@ -0,0 +1,9 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 3.0,
|
| 3 |
+
"total_flos": 6937506539055684.0,
|
| 4 |
+
"train_loss": 0.2534543433457448,
|
| 5 |
+
"train_runtime": 276.4842,
|
| 6 |
+
"train_samples": 67349,
|
| 7 |
+
"train_samples_per_second": 730.772,
|
| 8 |
+
"train_steps_per_second": 9.136
|
| 9 |
+
}
|
mbert_paper_metrics/result_ablation_mbert_paper/sst2/mBERT/gated_multi_branch/seed_44/trainer_state.json
ADDED
|
@@ -0,0 +1,419 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_metric": null,
|
| 3 |
+
"best_model_checkpoint": null,
|
| 4 |
+
"epoch": 3.0,
|
| 5 |
+
"eval_steps": 500,
|
| 6 |
+
"global_step": 2526,
|
| 7 |
+
"is_hyper_param_search": false,
|
| 8 |
+
"is_local_process_zero": true,
|
| 9 |
+
"is_world_process_zero": true,
|
| 10 |
+
"log_history": [
|
| 11 |
+
{
|
| 12 |
+
"epoch": 0.05938242280285035,
|
| 13 |
+
"grad_norm": 4.468812942504883,
|
| 14 |
+
"learning_rate": 1.9604117181314332e-05,
|
| 15 |
+
"loss": 0.564,
|
| 16 |
+
"step": 50
|
| 17 |
+
},
|
| 18 |
+
{
|
| 19 |
+
"epoch": 0.1187648456057007,
|
| 20 |
+
"grad_norm": 3.6407744884490967,
|
| 21 |
+
"learning_rate": 1.9208234362628665e-05,
|
| 22 |
+
"loss": 0.4508,
|
| 23 |
+
"step": 100
|
| 24 |
+
},
|
| 25 |
+
{
|
| 26 |
+
"epoch": 0.17814726840855108,
|
| 27 |
+
"grad_norm": 5.858651638031006,
|
| 28 |
+
"learning_rate": 1.8812351543942996e-05,
|
| 29 |
+
"loss": 0.426,
|
| 30 |
+
"step": 150
|
| 31 |
+
},
|
| 32 |
+
{
|
| 33 |
+
"epoch": 0.2375296912114014,
|
| 34 |
+
"grad_norm": 3.9081571102142334,
|
| 35 |
+
"learning_rate": 1.8416468725257326e-05,
|
| 36 |
+
"loss": 0.3875,
|
| 37 |
+
"step": 200
|
| 38 |
+
},
|
| 39 |
+
{
|
| 40 |
+
"epoch": 0.29691211401425177,
|
| 41 |
+
"grad_norm": 4.128037929534912,
|
| 42 |
+
"learning_rate": 1.8020585906571656e-05,
|
| 43 |
+
"loss": 0.3895,
|
| 44 |
+
"step": 250
|
| 45 |
+
},
|
| 46 |
+
{
|
| 47 |
+
"epoch": 0.35629453681710216,
|
| 48 |
+
"grad_norm": 6.987155914306641,
|
| 49 |
+
"learning_rate": 1.762470308788599e-05,
|
| 50 |
+
"loss": 0.3847,
|
| 51 |
+
"step": 300
|
| 52 |
+
},
|
| 53 |
+
{
|
| 54 |
+
"epoch": 0.4156769596199525,
|
| 55 |
+
"grad_norm": 3.920654535293579,
|
| 56 |
+
"learning_rate": 1.7228820269200316e-05,
|
| 57 |
+
"loss": 0.3669,
|
| 58 |
+
"step": 350
|
| 59 |
+
},
|
| 60 |
+
{
|
| 61 |
+
"epoch": 0.4750593824228028,
|
| 62 |
+
"grad_norm": 4.469102382659912,
|
| 63 |
+
"learning_rate": 1.6832937450514647e-05,
|
| 64 |
+
"loss": 0.3458,
|
| 65 |
+
"step": 400
|
| 66 |
+
},
|
| 67 |
+
{
|
| 68 |
+
"epoch": 0.5344418052256532,
|
| 69 |
+
"grad_norm": 5.952022552490234,
|
| 70 |
+
"learning_rate": 1.643705463182898e-05,
|
| 71 |
+
"loss": 0.3254,
|
| 72 |
+
"step": 450
|
| 73 |
+
},
|
| 74 |
+
{
|
| 75 |
+
"epoch": 0.5938242280285035,
|
| 76 |
+
"grad_norm": 6.218475341796875,
|
| 77 |
+
"learning_rate": 1.604117181314331e-05,
|
| 78 |
+
"loss": 0.3356,
|
| 79 |
+
"step": 500
|
| 80 |
+
},
|
| 81 |
+
{
|
| 82 |
+
"epoch": 0.6532066508313539,
|
| 83 |
+
"grad_norm": 6.0158796310424805,
|
| 84 |
+
"learning_rate": 1.564528899445764e-05,
|
| 85 |
+
"loss": 0.3323,
|
| 86 |
+
"step": 550
|
| 87 |
+
},
|
| 88 |
+
{
|
| 89 |
+
"epoch": 0.7125890736342043,
|
| 90 |
+
"grad_norm": 6.247025489807129,
|
| 91 |
+
"learning_rate": 1.5249406175771972e-05,
|
| 92 |
+
"loss": 0.3162,
|
| 93 |
+
"step": 600
|
| 94 |
+
},
|
| 95 |
+
{
|
| 96 |
+
"epoch": 0.7719714964370546,
|
| 97 |
+
"grad_norm": 5.0037689208984375,
|
| 98 |
+
"learning_rate": 1.4853523357086304e-05,
|
| 99 |
+
"loss": 0.3205,
|
| 100 |
+
"step": 650
|
| 101 |
+
},
|
| 102 |
+
{
|
| 103 |
+
"epoch": 0.831353919239905,
|
| 104 |
+
"grad_norm": 5.005596160888672,
|
| 105 |
+
"learning_rate": 1.4457640538400634e-05,
|
| 106 |
+
"loss": 0.3011,
|
| 107 |
+
"step": 700
|
| 108 |
+
},
|
| 109 |
+
{
|
| 110 |
+
"epoch": 0.8907363420427553,
|
| 111 |
+
"grad_norm": 6.862703323364258,
|
| 112 |
+
"learning_rate": 1.4061757719714966e-05,
|
| 113 |
+
"loss": 0.298,
|
| 114 |
+
"step": 750
|
| 115 |
+
},
|
| 116 |
+
{
|
| 117 |
+
"epoch": 0.9501187648456056,
|
| 118 |
+
"grad_norm": 6.259370803833008,
|
| 119 |
+
"learning_rate": 1.3665874901029297e-05,
|
| 120 |
+
"loss": 0.3026,
|
| 121 |
+
"step": 800
|
| 122 |
+
},
|
| 123 |
+
{
|
| 124 |
+
"epoch": 1.0,
|
| 125 |
+
"eval_accuracy": 0.856651376146789,
|
| 126 |
+
"eval_loss": 0.3314654231071472,
|
| 127 |
+
"eval_runtime": 2.2512,
|
| 128 |
+
"eval_samples_per_second": 387.342,
|
| 129 |
+
"eval_steps_per_second": 2.665,
|
| 130 |
+
"step": 842
|
| 131 |
+
},
|
| 132 |
+
{
|
| 133 |
+
"epoch": 1.009501187648456,
|
| 134 |
+
"grad_norm": 8.270345687866211,
|
| 135 |
+
"learning_rate": 1.3269992082343628e-05,
|
| 136 |
+
"loss": 0.2789,
|
| 137 |
+
"step": 850
|
| 138 |
+
},
|
| 139 |
+
{
|
| 140 |
+
"epoch": 1.0688836104513064,
|
| 141 |
+
"grad_norm": 7.798162937164307,
|
| 142 |
+
"learning_rate": 1.2874109263657959e-05,
|
| 143 |
+
"loss": 0.2507,
|
| 144 |
+
"step": 900
|
| 145 |
+
},
|
| 146 |
+
{
|
| 147 |
+
"epoch": 1.1282660332541568,
|
| 148 |
+
"grad_norm": 5.7158026695251465,
|
| 149 |
+
"learning_rate": 1.247822644497229e-05,
|
| 150 |
+
"loss": 0.2226,
|
| 151 |
+
"step": 950
|
| 152 |
+
},
|
| 153 |
+
{
|
| 154 |
+
"epoch": 1.187648456057007,
|
| 155 |
+
"grad_norm": 4.544774055480957,
|
| 156 |
+
"learning_rate": 1.208234362628662e-05,
|
| 157 |
+
"loss": 0.2362,
|
| 158 |
+
"step": 1000
|
| 159 |
+
},
|
| 160 |
+
{
|
| 161 |
+
"epoch": 1.2470308788598574,
|
| 162 |
+
"grad_norm": 6.844624996185303,
|
| 163 |
+
"learning_rate": 1.168646080760095e-05,
|
| 164 |
+
"loss": 0.2382,
|
| 165 |
+
"step": 1050
|
| 166 |
+
},
|
| 167 |
+
{
|
| 168 |
+
"epoch": 1.3064133016627077,
|
| 169 |
+
"grad_norm": 3.9297780990600586,
|
| 170 |
+
"learning_rate": 1.1290577988915281e-05,
|
| 171 |
+
"loss": 0.2344,
|
| 172 |
+
"step": 1100
|
| 173 |
+
},
|
| 174 |
+
{
|
| 175 |
+
"epoch": 1.365795724465558,
|
| 176 |
+
"grad_norm": 8.034307479858398,
|
| 177 |
+
"learning_rate": 1.0894695170229613e-05,
|
| 178 |
+
"loss": 0.2436,
|
| 179 |
+
"step": 1150
|
| 180 |
+
},
|
| 181 |
+
{
|
| 182 |
+
"epoch": 1.4251781472684084,
|
| 183 |
+
"grad_norm": 4.137219429016113,
|
| 184 |
+
"learning_rate": 1.0498812351543943e-05,
|
| 185 |
+
"loss": 0.2364,
|
| 186 |
+
"step": 1200
|
| 187 |
+
},
|
| 188 |
+
{
|
| 189 |
+
"epoch": 1.484560570071259,
|
| 190 |
+
"grad_norm": 4.177159786224365,
|
| 191 |
+
"learning_rate": 1.0102929532858275e-05,
|
| 192 |
+
"loss": 0.2307,
|
| 193 |
+
"step": 1250
|
| 194 |
+
},
|
| 195 |
+
{
|
| 196 |
+
"epoch": 1.5439429928741093,
|
| 197 |
+
"grad_norm": 4.810873031616211,
|
| 198 |
+
"learning_rate": 9.707046714172605e-06,
|
| 199 |
+
"loss": 0.2403,
|
| 200 |
+
"step": 1300
|
| 201 |
+
},
|
| 202 |
+
{
|
| 203 |
+
"epoch": 1.6033254156769596,
|
| 204 |
+
"grad_norm": 5.518557548522949,
|
| 205 |
+
"learning_rate": 9.311163895486937e-06,
|
| 206 |
+
"loss": 0.2302,
|
| 207 |
+
"step": 1350
|
| 208 |
+
},
|
| 209 |
+
{
|
| 210 |
+
"epoch": 1.66270783847981,
|
| 211 |
+
"grad_norm": 4.627058029174805,
|
| 212 |
+
"learning_rate": 8.915281076801267e-06,
|
| 213 |
+
"loss": 0.2114,
|
| 214 |
+
"step": 1400
|
| 215 |
+
},
|
| 216 |
+
{
|
| 217 |
+
"epoch": 1.7220902612826603,
|
| 218 |
+
"grad_norm": 3.9440369606018066,
|
| 219 |
+
"learning_rate": 8.5193982581156e-06,
|
| 220 |
+
"loss": 0.2334,
|
| 221 |
+
"step": 1450
|
| 222 |
+
},
|
| 223 |
+
{
|
| 224 |
+
"epoch": 1.7814726840855108,
|
| 225 |
+
"grad_norm": 5.9406046867370605,
|
| 226 |
+
"learning_rate": 8.12351543942993e-06,
|
| 227 |
+
"loss": 0.2221,
|
| 228 |
+
"step": 1500
|
| 229 |
+
},
|
| 230 |
+
{
|
| 231 |
+
"epoch": 1.8408551068883612,
|
| 232 |
+
"grad_norm": 4.831118106842041,
|
| 233 |
+
"learning_rate": 7.72763262074426e-06,
|
| 234 |
+
"loss": 0.2215,
|
| 235 |
+
"step": 1550
|
| 236 |
+
},
|
| 237 |
+
{
|
| 238 |
+
"epoch": 1.9002375296912115,
|
| 239 |
+
"grad_norm": 6.496983528137207,
|
| 240 |
+
"learning_rate": 7.331749802058591e-06,
|
| 241 |
+
"loss": 0.2154,
|
| 242 |
+
"step": 1600
|
| 243 |
+
},
|
| 244 |
+
{
|
| 245 |
+
"epoch": 1.9596199524940618,
|
| 246 |
+
"grad_norm": 4.407467842102051,
|
| 247 |
+
"learning_rate": 6.935866983372922e-06,
|
| 248 |
+
"loss": 0.2263,
|
| 249 |
+
"step": 1650
|
| 250 |
+
},
|
| 251 |
+
{
|
| 252 |
+
"epoch": 2.0,
|
| 253 |
+
"eval_accuracy": 0.8795871559633027,
|
| 254 |
+
"eval_loss": 0.33930179476737976,
|
| 255 |
+
"eval_runtime": 2.1682,
|
| 256 |
+
"eval_samples_per_second": 402.171,
|
| 257 |
+
"eval_steps_per_second": 2.767,
|
| 258 |
+
"step": 1684
|
| 259 |
+
},
|
| 260 |
+
{
|
| 261 |
+
"epoch": 2.019002375296912,
|
| 262 |
+
"grad_norm": 4.340859889984131,
|
| 263 |
+
"learning_rate": 6.539984164687253e-06,
|
| 264 |
+
"loss": 0.1995,
|
| 265 |
+
"step": 1700
|
| 266 |
+
},
|
| 267 |
+
{
|
| 268 |
+
"epoch": 2.0783847980997625,
|
| 269 |
+
"grad_norm": 5.748497009277344,
|
| 270 |
+
"learning_rate": 6.1441013460015846e-06,
|
| 271 |
+
"loss": 0.1743,
|
| 272 |
+
"step": 1750
|
| 273 |
+
},
|
| 274 |
+
{
|
| 275 |
+
"epoch": 2.137767220902613,
|
| 276 |
+
"grad_norm": 6.349128246307373,
|
| 277 |
+
"learning_rate": 5.748218527315916e-06,
|
| 278 |
+
"loss": 0.16,
|
| 279 |
+
"step": 1800
|
| 280 |
+
},
|
| 281 |
+
{
|
| 282 |
+
"epoch": 2.197149643705463,
|
| 283 |
+
"grad_norm": 5.214664936065674,
|
| 284 |
+
"learning_rate": 5.352335708630247e-06,
|
| 285 |
+
"loss": 0.1749,
|
| 286 |
+
"step": 1850
|
| 287 |
+
},
|
| 288 |
+
{
|
| 289 |
+
"epoch": 2.2565320665083135,
|
| 290 |
+
"grad_norm": 6.889070510864258,
|
| 291 |
+
"learning_rate": 4.956452889944577e-06,
|
| 292 |
+
"loss": 0.1641,
|
| 293 |
+
"step": 1900
|
| 294 |
+
},
|
| 295 |
+
{
|
| 296 |
+
"epoch": 2.315914489311164,
|
| 297 |
+
"grad_norm": 6.826782703399658,
|
| 298 |
+
"learning_rate": 4.560570071258908e-06,
|
| 299 |
+
"loss": 0.1743,
|
| 300 |
+
"step": 1950
|
| 301 |
+
},
|
| 302 |
+
{
|
| 303 |
+
"epoch": 2.375296912114014,
|
| 304 |
+
"grad_norm": 4.2961201667785645,
|
| 305 |
+
"learning_rate": 4.164687252573238e-06,
|
| 306 |
+
"loss": 0.168,
|
| 307 |
+
"step": 2000
|
| 308 |
+
},
|
| 309 |
+
{
|
| 310 |
+
"epoch": 2.4346793349168645,
|
| 311 |
+
"grad_norm": 6.859589576721191,
|
| 312 |
+
"learning_rate": 3.7688044338875695e-06,
|
| 313 |
+
"loss": 0.1684,
|
| 314 |
+
"step": 2050
|
| 315 |
+
},
|
| 316 |
+
{
|
| 317 |
+
"epoch": 2.494061757719715,
|
| 318 |
+
"grad_norm": 4.703085899353027,
|
| 319 |
+
"learning_rate": 3.3729216152019006e-06,
|
| 320 |
+
"loss": 0.1627,
|
| 321 |
+
"step": 2100
|
| 322 |
+
},
|
| 323 |
+
{
|
| 324 |
+
"epoch": 2.553444180522565,
|
| 325 |
+
"grad_norm": 8.655251502990723,
|
| 326 |
+
"learning_rate": 2.977038796516231e-06,
|
| 327 |
+
"loss": 0.1645,
|
| 328 |
+
"step": 2150
|
| 329 |
+
},
|
| 330 |
+
{
|
| 331 |
+
"epoch": 2.6128266033254155,
|
| 332 |
+
"grad_norm": 5.534372806549072,
|
| 333 |
+
"learning_rate": 2.581155977830562e-06,
|
| 334 |
+
"loss": 0.1713,
|
| 335 |
+
"step": 2200
|
| 336 |
+
},
|
| 337 |
+
{
|
| 338 |
+
"epoch": 2.6722090261282663,
|
| 339 |
+
"grad_norm": 8.41250228881836,
|
| 340 |
+
"learning_rate": 2.1852731591448932e-06,
|
| 341 |
+
"loss": 0.1616,
|
| 342 |
+
"step": 2250
|
| 343 |
+
},
|
| 344 |
+
{
|
| 345 |
+
"epoch": 2.731591448931116,
|
| 346 |
+
"grad_norm": 4.01090145111084,
|
| 347 |
+
"learning_rate": 1.789390340459224e-06,
|
| 348 |
+
"loss": 0.1827,
|
| 349 |
+
"step": 2300
|
| 350 |
+
},
|
| 351 |
+
{
|
| 352 |
+
"epoch": 2.790973871733967,
|
| 353 |
+
"grad_norm": 4.952667713165283,
|
| 354 |
+
"learning_rate": 1.3935075217735553e-06,
|
| 355 |
+
"loss": 0.1753,
|
| 356 |
+
"step": 2350
|
| 357 |
+
},
|
| 358 |
+
{
|
| 359 |
+
"epoch": 2.850356294536817,
|
| 360 |
+
"grad_norm": 7.208967685699463,
|
| 361 |
+
"learning_rate": 9.976247030878861e-07,
|
| 362 |
+
"loss": 0.1724,
|
| 363 |
+
"step": 2400
|
| 364 |
+
},
|
| 365 |
+
{
|
| 366 |
+
"epoch": 2.9097387173396676,
|
| 367 |
+
"grad_norm": 5.433742523193359,
|
| 368 |
+
"learning_rate": 6.01741884402217e-07,
|
| 369 |
+
"loss": 0.1666,
|
| 370 |
+
"step": 2450
|
| 371 |
+
},
|
| 372 |
+
{
|
| 373 |
+
"epoch": 2.969121140142518,
|
| 374 |
+
"grad_norm": 4.888383388519287,
|
| 375 |
+
"learning_rate": 2.0585906571654793e-07,
|
| 376 |
+
"loss": 0.1604,
|
| 377 |
+
"step": 2500
|
| 378 |
+
},
|
| 379 |
+
{
|
| 380 |
+
"epoch": 3.0,
|
| 381 |
+
"eval_accuracy": 0.875,
|
| 382 |
+
"eval_loss": 0.3715299665927887,
|
| 383 |
+
"eval_runtime": 2.0675,
|
| 384 |
+
"eval_samples_per_second": 421.772,
|
| 385 |
+
"eval_steps_per_second": 2.902,
|
| 386 |
+
"step": 2526
|
| 387 |
+
},
|
| 388 |
+
{
|
| 389 |
+
"epoch": 3.0,
|
| 390 |
+
"step": 2526,
|
| 391 |
+
"total_flos": 6937506539055684.0,
|
| 392 |
+
"train_loss": 0.2534543433457448,
|
| 393 |
+
"train_runtime": 276.4842,
|
| 394 |
+
"train_samples_per_second": 730.772,
|
| 395 |
+
"train_steps_per_second": 9.136
|
| 396 |
+
}
|
| 397 |
+
],
|
| 398 |
+
"logging_steps": 50,
|
| 399 |
+
"max_steps": 2526,
|
| 400 |
+
"num_input_tokens_seen": 0,
|
| 401 |
+
"num_train_epochs": 3,
|
| 402 |
+
"save_steps": 500,
|
| 403 |
+
"stateful_callbacks": {
|
| 404 |
+
"TrainerControl": {
|
| 405 |
+
"args": {
|
| 406 |
+
"should_epoch_stop": false,
|
| 407 |
+
"should_evaluate": false,
|
| 408 |
+
"should_log": false,
|
| 409 |
+
"should_save": false,
|
| 410 |
+
"should_training_stop": false
|
| 411 |
+
},
|
| 412 |
+
"attributes": {}
|
| 413 |
+
}
|
| 414 |
+
},
|
| 415 |
+
"total_flos": 6937506539055684.0,
|
| 416 |
+
"train_batch_size": 80,
|
| 417 |
+
"trial_name": null,
|
| 418 |
+
"trial_params": null
|
| 419 |
+
}
|
mbert_paper_metrics/result_ablation_mbert_paper/sst2/mBERT/hf_sequence_classifier/seed_42/all_results.json
ADDED
|
@@ -0,0 +1,15 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 3.0,
|
| 3 |
+
"eval_accuracy": 0.8692660550458715,
|
| 4 |
+
"eval_loss": 0.41583138704299927,
|
| 5 |
+
"eval_runtime": 2.4702,
|
| 6 |
+
"eval_samples": 872,
|
| 7 |
+
"eval_samples_per_second": 353.002,
|
| 8 |
+
"eval_steps_per_second": 2.429,
|
| 9 |
+
"total_flos": 6601310434568880.0,
|
| 10 |
+
"train_loss": 0.25817311102685153,
|
| 11 |
+
"train_runtime": 249.3454,
|
| 12 |
+
"train_samples": 67349,
|
| 13 |
+
"train_samples_per_second": 810.31,
|
| 14 |
+
"train_steps_per_second": 10.131
|
| 15 |
+
}
|
mbert_paper_metrics/result_ablation_mbert_paper/sst2/mBERT/hf_sequence_classifier/seed_42/eval_results.json
ADDED
|
@@ -0,0 +1,9 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 3.0,
|
| 3 |
+
"eval_accuracy": 0.8692660550458715,
|
| 4 |
+
"eval_loss": 0.41583138704299927,
|
| 5 |
+
"eval_runtime": 2.4702,
|
| 6 |
+
"eval_samples": 872,
|
| 7 |
+
"eval_samples_per_second": 353.002,
|
| 8 |
+
"eval_steps_per_second": 2.429
|
| 9 |
+
}
|
mbert_paper_metrics/result_ablation_mbert_paper/sst2/mBERT/hf_sequence_classifier/seed_42/run_command.txt
ADDED
|
@@ -0,0 +1 @@
|
|
|
|
|
|
|
| 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 sst2 --max_seq_length 256 --do_train --do_eval --evaluation_strategy epoch --save_strategy no --logging_strategy steps --logging_steps 50 --report_to none --num_train_epochs 3 --learning_rate 2e-05 --per_device_train_batch_size 80 --per_device_eval_batch_size 168 --gradient_accumulation_steps 1 --dataloader_num_workers 2 --seed 42 --output_dir result_ablation_mbert_paper/sst2/mBERT/hf_sequence_classifier/seed_42 --overwrite_output_dir --trust_remote_code true --bf16 --tf32 true --skip_model_save
|
mbert_paper_metrics/result_ablation_mbert_paper/sst2/mBERT/hf_sequence_classifier/seed_42/train_results.json
ADDED
|
@@ -0,0 +1,9 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 3.0,
|
| 3 |
+
"total_flos": 6601310434568880.0,
|
| 4 |
+
"train_loss": 0.25817311102685153,
|
| 5 |
+
"train_runtime": 249.3454,
|
| 6 |
+
"train_samples": 67349,
|
| 7 |
+
"train_samples_per_second": 810.31,
|
| 8 |
+
"train_steps_per_second": 10.131
|
| 9 |
+
}
|
mbert_paper_metrics/result_ablation_mbert_paper/sst2/mBERT/hf_sequence_classifier/seed_42/trainer_state.json
ADDED
|
@@ -0,0 +1,419 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_metric": null,
|
| 3 |
+
"best_model_checkpoint": null,
|
| 4 |
+
"epoch": 3.0,
|
| 5 |
+
"eval_steps": 500,
|
| 6 |
+
"global_step": 2526,
|
| 7 |
+
"is_hyper_param_search": false,
|
| 8 |
+
"is_local_process_zero": true,
|
| 9 |
+
"is_world_process_zero": true,
|
| 10 |
+
"log_history": [
|
| 11 |
+
{
|
| 12 |
+
"epoch": 0.05938242280285035,
|
| 13 |
+
"grad_norm": 5.259804725646973,
|
| 14 |
+
"learning_rate": 1.9604117181314332e-05,
|
| 15 |
+
"loss": 0.5922,
|
| 16 |
+
"step": 50
|
| 17 |
+
},
|
| 18 |
+
{
|
| 19 |
+
"epoch": 0.1187648456057007,
|
| 20 |
+
"grad_norm": 4.923093795776367,
|
| 21 |
+
"learning_rate": 1.9208234362628665e-05,
|
| 22 |
+
"loss": 0.4774,
|
| 23 |
+
"step": 100
|
| 24 |
+
},
|
| 25 |
+
{
|
| 26 |
+
"epoch": 0.17814726840855108,
|
| 27 |
+
"grad_norm": 5.227987289428711,
|
| 28 |
+
"learning_rate": 1.8812351543942996e-05,
|
| 29 |
+
"loss": 0.4239,
|
| 30 |
+
"step": 150
|
| 31 |
+
},
|
| 32 |
+
{
|
| 33 |
+
"epoch": 0.2375296912114014,
|
| 34 |
+
"grad_norm": 4.725414752960205,
|
| 35 |
+
"learning_rate": 1.8416468725257326e-05,
|
| 36 |
+
"loss": 0.4159,
|
| 37 |
+
"step": 200
|
| 38 |
+
},
|
| 39 |
+
{
|
| 40 |
+
"epoch": 0.29691211401425177,
|
| 41 |
+
"grad_norm": 6.134373664855957,
|
| 42 |
+
"learning_rate": 1.8020585906571656e-05,
|
| 43 |
+
"loss": 0.3937,
|
| 44 |
+
"step": 250
|
| 45 |
+
},
|
| 46 |
+
{
|
| 47 |
+
"epoch": 0.35629453681710216,
|
| 48 |
+
"grad_norm": 4.437064170837402,
|
| 49 |
+
"learning_rate": 1.762470308788599e-05,
|
| 50 |
+
"loss": 0.3838,
|
| 51 |
+
"step": 300
|
| 52 |
+
},
|
| 53 |
+
{
|
| 54 |
+
"epoch": 0.4156769596199525,
|
| 55 |
+
"grad_norm": 3.7955241203308105,
|
| 56 |
+
"learning_rate": 1.7228820269200316e-05,
|
| 57 |
+
"loss": 0.3525,
|
| 58 |
+
"step": 350
|
| 59 |
+
},
|
| 60 |
+
{
|
| 61 |
+
"epoch": 0.4750593824228028,
|
| 62 |
+
"grad_norm": 5.348527431488037,
|
| 63 |
+
"learning_rate": 1.6832937450514647e-05,
|
| 64 |
+
"loss": 0.3477,
|
| 65 |
+
"step": 400
|
| 66 |
+
},
|
| 67 |
+
{
|
| 68 |
+
"epoch": 0.5344418052256532,
|
| 69 |
+
"grad_norm": 5.955068111419678,
|
| 70 |
+
"learning_rate": 1.643705463182898e-05,
|
| 71 |
+
"loss": 0.3444,
|
| 72 |
+
"step": 450
|
| 73 |
+
},
|
| 74 |
+
{
|
| 75 |
+
"epoch": 0.5938242280285035,
|
| 76 |
+
"grad_norm": 4.143620491027832,
|
| 77 |
+
"learning_rate": 1.604117181314331e-05,
|
| 78 |
+
"loss": 0.3314,
|
| 79 |
+
"step": 500
|
| 80 |
+
},
|
| 81 |
+
{
|
| 82 |
+
"epoch": 0.6532066508313539,
|
| 83 |
+
"grad_norm": 4.2212114334106445,
|
| 84 |
+
"learning_rate": 1.564528899445764e-05,
|
| 85 |
+
"loss": 0.3277,
|
| 86 |
+
"step": 550
|
| 87 |
+
},
|
| 88 |
+
{
|
| 89 |
+
"epoch": 0.7125890736342043,
|
| 90 |
+
"grad_norm": 4.133213043212891,
|
| 91 |
+
"learning_rate": 1.5249406175771972e-05,
|
| 92 |
+
"loss": 0.3338,
|
| 93 |
+
"step": 600
|
| 94 |
+
},
|
| 95 |
+
{
|
| 96 |
+
"epoch": 0.7719714964370546,
|
| 97 |
+
"grad_norm": 4.429179668426514,
|
| 98 |
+
"learning_rate": 1.4853523357086304e-05,
|
| 99 |
+
"loss": 0.3121,
|
| 100 |
+
"step": 650
|
| 101 |
+
},
|
| 102 |
+
{
|
| 103 |
+
"epoch": 0.831353919239905,
|
| 104 |
+
"grad_norm": 4.174455165863037,
|
| 105 |
+
"learning_rate": 1.4457640538400634e-05,
|
| 106 |
+
"loss": 0.3161,
|
| 107 |
+
"step": 700
|
| 108 |
+
},
|
| 109 |
+
{
|
| 110 |
+
"epoch": 0.8907363420427553,
|
| 111 |
+
"grad_norm": 4.580986499786377,
|
| 112 |
+
"learning_rate": 1.4061757719714966e-05,
|
| 113 |
+
"loss": 0.3044,
|
| 114 |
+
"step": 750
|
| 115 |
+
},
|
| 116 |
+
{
|
| 117 |
+
"epoch": 0.9501187648456056,
|
| 118 |
+
"grad_norm": 4.160976409912109,
|
| 119 |
+
"learning_rate": 1.3665874901029297e-05,
|
| 120 |
+
"loss": 0.2905,
|
| 121 |
+
"step": 800
|
| 122 |
+
},
|
| 123 |
+
{
|
| 124 |
+
"epoch": 1.0,
|
| 125 |
+
"eval_accuracy": 0.8543577981651376,
|
| 126 |
+
"eval_loss": 0.35903483629226685,
|
| 127 |
+
"eval_runtime": 2.4644,
|
| 128 |
+
"eval_samples_per_second": 353.84,
|
| 129 |
+
"eval_steps_per_second": 2.435,
|
| 130 |
+
"step": 842
|
| 131 |
+
},
|
| 132 |
+
{
|
| 133 |
+
"epoch": 1.009501187648456,
|
| 134 |
+
"grad_norm": 3.4890060424804688,
|
| 135 |
+
"learning_rate": 1.3269992082343628e-05,
|
| 136 |
+
"loss": 0.3013,
|
| 137 |
+
"step": 850
|
| 138 |
+
},
|
| 139 |
+
{
|
| 140 |
+
"epoch": 1.0688836104513064,
|
| 141 |
+
"grad_norm": 4.328893661499023,
|
| 142 |
+
"learning_rate": 1.2874109263657959e-05,
|
| 143 |
+
"loss": 0.2633,
|
| 144 |
+
"step": 900
|
| 145 |
+
},
|
| 146 |
+
{
|
| 147 |
+
"epoch": 1.1282660332541568,
|
| 148 |
+
"grad_norm": 3.6598289012908936,
|
| 149 |
+
"learning_rate": 1.247822644497229e-05,
|
| 150 |
+
"loss": 0.2441,
|
| 151 |
+
"step": 950
|
| 152 |
+
},
|
| 153 |
+
{
|
| 154 |
+
"epoch": 1.187648456057007,
|
| 155 |
+
"grad_norm": 5.306523323059082,
|
| 156 |
+
"learning_rate": 1.208234362628662e-05,
|
| 157 |
+
"loss": 0.2361,
|
| 158 |
+
"step": 1000
|
| 159 |
+
},
|
| 160 |
+
{
|
| 161 |
+
"epoch": 1.2470308788598574,
|
| 162 |
+
"grad_norm": 5.256509304046631,
|
| 163 |
+
"learning_rate": 1.168646080760095e-05,
|
| 164 |
+
"loss": 0.2617,
|
| 165 |
+
"step": 1050
|
| 166 |
+
},
|
| 167 |
+
{
|
| 168 |
+
"epoch": 1.3064133016627077,
|
| 169 |
+
"grad_norm": 4.3957343101501465,
|
| 170 |
+
"learning_rate": 1.1290577988915281e-05,
|
| 171 |
+
"loss": 0.2422,
|
| 172 |
+
"step": 1100
|
| 173 |
+
},
|
| 174 |
+
{
|
| 175 |
+
"epoch": 1.365795724465558,
|
| 176 |
+
"grad_norm": 7.06696081161499,
|
| 177 |
+
"learning_rate": 1.0894695170229613e-05,
|
| 178 |
+
"loss": 0.2229,
|
| 179 |
+
"step": 1150
|
| 180 |
+
},
|
| 181 |
+
{
|
| 182 |
+
"epoch": 1.4251781472684084,
|
| 183 |
+
"grad_norm": 5.169172763824463,
|
| 184 |
+
"learning_rate": 1.0498812351543943e-05,
|
| 185 |
+
"loss": 0.2295,
|
| 186 |
+
"step": 1200
|
| 187 |
+
},
|
| 188 |
+
{
|
| 189 |
+
"epoch": 1.484560570071259,
|
| 190 |
+
"grad_norm": 7.6731977462768555,
|
| 191 |
+
"learning_rate": 1.0102929532858275e-05,
|
| 192 |
+
"loss": 0.2393,
|
| 193 |
+
"step": 1250
|
| 194 |
+
},
|
| 195 |
+
{
|
| 196 |
+
"epoch": 1.5439429928741093,
|
| 197 |
+
"grad_norm": 6.629937171936035,
|
| 198 |
+
"learning_rate": 9.707046714172605e-06,
|
| 199 |
+
"loss": 0.2431,
|
| 200 |
+
"step": 1300
|
| 201 |
+
},
|
| 202 |
+
{
|
| 203 |
+
"epoch": 1.6033254156769596,
|
| 204 |
+
"grad_norm": 8.116971015930176,
|
| 205 |
+
"learning_rate": 9.311163895486937e-06,
|
| 206 |
+
"loss": 0.2057,
|
| 207 |
+
"step": 1350
|
| 208 |
+
},
|
| 209 |
+
{
|
| 210 |
+
"epoch": 1.66270783847981,
|
| 211 |
+
"grad_norm": 3.551285982131958,
|
| 212 |
+
"learning_rate": 8.915281076801267e-06,
|
| 213 |
+
"loss": 0.2496,
|
| 214 |
+
"step": 1400
|
| 215 |
+
},
|
| 216 |
+
{
|
| 217 |
+
"epoch": 1.7220902612826603,
|
| 218 |
+
"grad_norm": 3.743298053741455,
|
| 219 |
+
"learning_rate": 8.5193982581156e-06,
|
| 220 |
+
"loss": 0.2102,
|
| 221 |
+
"step": 1450
|
| 222 |
+
},
|
| 223 |
+
{
|
| 224 |
+
"epoch": 1.7814726840855108,
|
| 225 |
+
"grad_norm": 5.7800822257995605,
|
| 226 |
+
"learning_rate": 8.12351543942993e-06,
|
| 227 |
+
"loss": 0.215,
|
| 228 |
+
"step": 1500
|
| 229 |
+
},
|
| 230 |
+
{
|
| 231 |
+
"epoch": 1.8408551068883612,
|
| 232 |
+
"grad_norm": 6.816977024078369,
|
| 233 |
+
"learning_rate": 7.72763262074426e-06,
|
| 234 |
+
"loss": 0.2301,
|
| 235 |
+
"step": 1550
|
| 236 |
+
},
|
| 237 |
+
{
|
| 238 |
+
"epoch": 1.9002375296912115,
|
| 239 |
+
"grad_norm": 4.686933517456055,
|
| 240 |
+
"learning_rate": 7.331749802058591e-06,
|
| 241 |
+
"loss": 0.2193,
|
| 242 |
+
"step": 1600
|
| 243 |
+
},
|
| 244 |
+
{
|
| 245 |
+
"epoch": 1.9596199524940618,
|
| 246 |
+
"grad_norm": 9.363842964172363,
|
| 247 |
+
"learning_rate": 6.935866983372922e-06,
|
| 248 |
+
"loss": 0.2179,
|
| 249 |
+
"step": 1650
|
| 250 |
+
},
|
| 251 |
+
{
|
| 252 |
+
"epoch": 2.0,
|
| 253 |
+
"eval_accuracy": 0.8635321100917431,
|
| 254 |
+
"eval_loss": 0.3860330879688263,
|
| 255 |
+
"eval_runtime": 2.5125,
|
| 256 |
+
"eval_samples_per_second": 347.061,
|
| 257 |
+
"eval_steps_per_second": 2.388,
|
| 258 |
+
"step": 1684
|
| 259 |
+
},
|
| 260 |
+
{
|
| 261 |
+
"epoch": 2.019002375296912,
|
| 262 |
+
"grad_norm": 6.171146392822266,
|
| 263 |
+
"learning_rate": 6.539984164687253e-06,
|
| 264 |
+
"loss": 0.2032,
|
| 265 |
+
"step": 1700
|
| 266 |
+
},
|
| 267 |
+
{
|
| 268 |
+
"epoch": 2.0783847980997625,
|
| 269 |
+
"grad_norm": 5.172806262969971,
|
| 270 |
+
"learning_rate": 6.1441013460015846e-06,
|
| 271 |
+
"loss": 0.1747,
|
| 272 |
+
"step": 1750
|
| 273 |
+
},
|
| 274 |
+
{
|
| 275 |
+
"epoch": 2.137767220902613,
|
| 276 |
+
"grad_norm": 4.122064590454102,
|
| 277 |
+
"learning_rate": 5.748218527315916e-06,
|
| 278 |
+
"loss": 0.1745,
|
| 279 |
+
"step": 1800
|
| 280 |
+
},
|
| 281 |
+
{
|
| 282 |
+
"epoch": 2.197149643705463,
|
| 283 |
+
"grad_norm": 6.982349395751953,
|
| 284 |
+
"learning_rate": 5.352335708630247e-06,
|
| 285 |
+
"loss": 0.1787,
|
| 286 |
+
"step": 1850
|
| 287 |
+
},
|
| 288 |
+
{
|
| 289 |
+
"epoch": 2.2565320665083135,
|
| 290 |
+
"grad_norm": 7.98695707321167,
|
| 291 |
+
"learning_rate": 4.956452889944577e-06,
|
| 292 |
+
"loss": 0.1815,
|
| 293 |
+
"step": 1900
|
| 294 |
+
},
|
| 295 |
+
{
|
| 296 |
+
"epoch": 2.315914489311164,
|
| 297 |
+
"grad_norm": 5.30454158782959,
|
| 298 |
+
"learning_rate": 4.560570071258908e-06,
|
| 299 |
+
"loss": 0.1772,
|
| 300 |
+
"step": 1950
|
| 301 |
+
},
|
| 302 |
+
{
|
| 303 |
+
"epoch": 2.375296912114014,
|
| 304 |
+
"grad_norm": 5.302230358123779,
|
| 305 |
+
"learning_rate": 4.164687252573238e-06,
|
| 306 |
+
"loss": 0.1834,
|
| 307 |
+
"step": 2000
|
| 308 |
+
},
|
| 309 |
+
{
|
| 310 |
+
"epoch": 2.4346793349168645,
|
| 311 |
+
"grad_norm": 5.417513847351074,
|
| 312 |
+
"learning_rate": 3.7688044338875695e-06,
|
| 313 |
+
"loss": 0.174,
|
| 314 |
+
"step": 2050
|
| 315 |
+
},
|
| 316 |
+
{
|
| 317 |
+
"epoch": 2.494061757719715,
|
| 318 |
+
"grad_norm": 5.009998798370361,
|
| 319 |
+
"learning_rate": 3.3729216152019006e-06,
|
| 320 |
+
"loss": 0.1546,
|
| 321 |
+
"step": 2100
|
| 322 |
+
},
|
| 323 |
+
{
|
| 324 |
+
"epoch": 2.553444180522565,
|
| 325 |
+
"grad_norm": 4.281483173370361,
|
| 326 |
+
"learning_rate": 2.977038796516231e-06,
|
| 327 |
+
"loss": 0.1782,
|
| 328 |
+
"step": 2150
|
| 329 |
+
},
|
| 330 |
+
{
|
| 331 |
+
"epoch": 2.6128266033254155,
|
| 332 |
+
"grad_norm": 6.131183624267578,
|
| 333 |
+
"learning_rate": 2.581155977830562e-06,
|
| 334 |
+
"loss": 0.1839,
|
| 335 |
+
"step": 2200
|
| 336 |
+
},
|
| 337 |
+
{
|
| 338 |
+
"epoch": 2.6722090261282663,
|
| 339 |
+
"grad_norm": 4.814934253692627,
|
| 340 |
+
"learning_rate": 2.1852731591448932e-06,
|
| 341 |
+
"loss": 0.1756,
|
| 342 |
+
"step": 2250
|
| 343 |
+
},
|
| 344 |
+
{
|
| 345 |
+
"epoch": 2.731591448931116,
|
| 346 |
+
"grad_norm": 8.870471000671387,
|
| 347 |
+
"learning_rate": 1.789390340459224e-06,
|
| 348 |
+
"loss": 0.1605,
|
| 349 |
+
"step": 2300
|
| 350 |
+
},
|
| 351 |
+
{
|
| 352 |
+
"epoch": 2.790973871733967,
|
| 353 |
+
"grad_norm": 6.901409149169922,
|
| 354 |
+
"learning_rate": 1.3935075217735553e-06,
|
| 355 |
+
"loss": 0.1779,
|
| 356 |
+
"step": 2350
|
| 357 |
+
},
|
| 358 |
+
{
|
| 359 |
+
"epoch": 2.850356294536817,
|
| 360 |
+
"grad_norm": 6.033817768096924,
|
| 361 |
+
"learning_rate": 9.976247030878861e-07,
|
| 362 |
+
"loss": 0.1744,
|
| 363 |
+
"step": 2400
|
| 364 |
+
},
|
| 365 |
+
{
|
| 366 |
+
"epoch": 2.9097387173396676,
|
| 367 |
+
"grad_norm": 7.492908477783203,
|
| 368 |
+
"learning_rate": 6.01741884402217e-07,
|
| 369 |
+
"loss": 0.1575,
|
| 370 |
+
"step": 2450
|
| 371 |
+
},
|
| 372 |
+
{
|
| 373 |
+
"epoch": 2.969121140142518,
|
| 374 |
+
"grad_norm": 6.021579265594482,
|
| 375 |
+
"learning_rate": 2.0585906571654793e-07,
|
| 376 |
+
"loss": 0.1685,
|
| 377 |
+
"step": 2500
|
| 378 |
+
},
|
| 379 |
+
{
|
| 380 |
+
"epoch": 3.0,
|
| 381 |
+
"eval_accuracy": 0.8692660550458715,
|
| 382 |
+
"eval_loss": 0.41583138704299927,
|
| 383 |
+
"eval_runtime": 2.4374,
|
| 384 |
+
"eval_samples_per_second": 357.765,
|
| 385 |
+
"eval_steps_per_second": 2.462,
|
| 386 |
+
"step": 2526
|
| 387 |
+
},
|
| 388 |
+
{
|
| 389 |
+
"epoch": 3.0,
|
| 390 |
+
"step": 2526,
|
| 391 |
+
"total_flos": 6601310434568880.0,
|
| 392 |
+
"train_loss": 0.25817311102685153,
|
| 393 |
+
"train_runtime": 249.3454,
|
| 394 |
+
"train_samples_per_second": 810.31,
|
| 395 |
+
"train_steps_per_second": 10.131
|
| 396 |
+
}
|
| 397 |
+
],
|
| 398 |
+
"logging_steps": 50,
|
| 399 |
+
"max_steps": 2526,
|
| 400 |
+
"num_input_tokens_seen": 0,
|
| 401 |
+
"num_train_epochs": 3,
|
| 402 |
+
"save_steps": 500,
|
| 403 |
+
"stateful_callbacks": {
|
| 404 |
+
"TrainerControl": {
|
| 405 |
+
"args": {
|
| 406 |
+
"should_epoch_stop": false,
|
| 407 |
+
"should_evaluate": false,
|
| 408 |
+
"should_log": false,
|
| 409 |
+
"should_save": false,
|
| 410 |
+
"should_training_stop": false
|
| 411 |
+
},
|
| 412 |
+
"attributes": {}
|
| 413 |
+
}
|
| 414 |
+
},
|
| 415 |
+
"total_flos": 6601310434568880.0,
|
| 416 |
+
"train_batch_size": 80,
|
| 417 |
+
"trial_name": null,
|
| 418 |
+
"trial_params": null
|
| 419 |
+
}
|
mbert_paper_metrics/result_ablation_mbert_paper/sst2/mBERT/hf_sequence_classifier/seed_43/all_results.json
ADDED
|
@@ -0,0 +1,15 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 3.0,
|
| 3 |
+
"eval_accuracy": 0.8784403669724771,
|
| 4 |
+
"eval_loss": 0.3952513635158539,
|
| 5 |
+
"eval_runtime": 2.3794,
|
| 6 |
+
"eval_samples": 872,
|
| 7 |
+
"eval_samples_per_second": 366.484,
|
| 8 |
+
"eval_steps_per_second": 2.522,
|
| 9 |
+
"total_flos": 6606837822286560.0,
|
| 10 |
+
"train_loss": 0.2572364091684397,
|
| 11 |
+
"train_runtime": 266.5223,
|
| 12 |
+
"train_samples": 67349,
|
| 13 |
+
"train_samples_per_second": 758.087,
|
| 14 |
+
"train_steps_per_second": 9.478
|
| 15 |
+
}
|
mbert_paper_metrics/result_ablation_mbert_paper/sst2/mBERT/hf_sequence_classifier/seed_43/eval_results.json
ADDED
|
@@ -0,0 +1,9 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 3.0,
|
| 3 |
+
"eval_accuracy": 0.8784403669724771,
|
| 4 |
+
"eval_loss": 0.3952513635158539,
|
| 5 |
+
"eval_runtime": 2.3794,
|
| 6 |
+
"eval_samples": 872,
|
| 7 |
+
"eval_samples_per_second": 366.484,
|
| 8 |
+
"eval_steps_per_second": 2.522
|
| 9 |
+
}
|
mbert_paper_metrics/result_ablation_mbert_paper/sst2/mBERT/hf_sequence_classifier/seed_43/run_command.txt
ADDED
|
@@ -0,0 +1 @@
|
|
|
|
|
|
|
| 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 sst2 --max_seq_length 256 --do_train --do_eval --evaluation_strategy epoch --save_strategy no --logging_strategy steps --logging_steps 50 --report_to none --num_train_epochs 3 --learning_rate 2e-05 --per_device_train_batch_size 80 --per_device_eval_batch_size 168 --gradient_accumulation_steps 1 --dataloader_num_workers 2 --seed 43 --output_dir result_ablation_mbert_paper/sst2/mBERT/hf_sequence_classifier/seed_43 --overwrite_output_dir --trust_remote_code true --bf16 --tf32 true --skip_model_save
|
mbert_paper_metrics/result_ablation_mbert_paper/sst2/mBERT/hf_sequence_classifier/seed_43/train_results.json
ADDED
|
@@ -0,0 +1,9 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 3.0,
|
| 3 |
+
"total_flos": 6606837822286560.0,
|
| 4 |
+
"train_loss": 0.2572364091684397,
|
| 5 |
+
"train_runtime": 266.5223,
|
| 6 |
+
"train_samples": 67349,
|
| 7 |
+
"train_samples_per_second": 758.087,
|
| 8 |
+
"train_steps_per_second": 9.478
|
| 9 |
+
}
|
mbert_paper_metrics/result_ablation_mbert_paper/sst2/mBERT/hf_sequence_classifier/seed_43/trainer_state.json
ADDED
|
@@ -0,0 +1,419 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_metric": null,
|
| 3 |
+
"best_model_checkpoint": null,
|
| 4 |
+
"epoch": 3.0,
|
| 5 |
+
"eval_steps": 500,
|
| 6 |
+
"global_step": 2526,
|
| 7 |
+
"is_hyper_param_search": false,
|
| 8 |
+
"is_local_process_zero": true,
|
| 9 |
+
"is_world_process_zero": true,
|
| 10 |
+
"log_history": [
|
| 11 |
+
{
|
| 12 |
+
"epoch": 0.05938242280285035,
|
| 13 |
+
"grad_norm": 5.042938709259033,
|
| 14 |
+
"learning_rate": 1.9604117181314332e-05,
|
| 15 |
+
"loss": 0.5941,
|
| 16 |
+
"step": 50
|
| 17 |
+
},
|
| 18 |
+
{
|
| 19 |
+
"epoch": 0.1187648456057007,
|
| 20 |
+
"grad_norm": 6.073488235473633,
|
| 21 |
+
"learning_rate": 1.9208234362628665e-05,
|
| 22 |
+
"loss": 0.4682,
|
| 23 |
+
"step": 100
|
| 24 |
+
},
|
| 25 |
+
{
|
| 26 |
+
"epoch": 0.17814726840855108,
|
| 27 |
+
"grad_norm": 5.6413044929504395,
|
| 28 |
+
"learning_rate": 1.8812351543942996e-05,
|
| 29 |
+
"loss": 0.4305,
|
| 30 |
+
"step": 150
|
| 31 |
+
},
|
| 32 |
+
{
|
| 33 |
+
"epoch": 0.2375296912114014,
|
| 34 |
+
"grad_norm": 5.005921840667725,
|
| 35 |
+
"learning_rate": 1.8416468725257326e-05,
|
| 36 |
+
"loss": 0.428,
|
| 37 |
+
"step": 200
|
| 38 |
+
},
|
| 39 |
+
{
|
| 40 |
+
"epoch": 0.29691211401425177,
|
| 41 |
+
"grad_norm": 4.58496618270874,
|
| 42 |
+
"learning_rate": 1.8020585906571656e-05,
|
| 43 |
+
"loss": 0.409,
|
| 44 |
+
"step": 250
|
| 45 |
+
},
|
| 46 |
+
{
|
| 47 |
+
"epoch": 0.35629453681710216,
|
| 48 |
+
"grad_norm": 3.5914125442504883,
|
| 49 |
+
"learning_rate": 1.762470308788599e-05,
|
| 50 |
+
"loss": 0.3786,
|
| 51 |
+
"step": 300
|
| 52 |
+
},
|
| 53 |
+
{
|
| 54 |
+
"epoch": 0.4156769596199525,
|
| 55 |
+
"grad_norm": 5.395323276519775,
|
| 56 |
+
"learning_rate": 1.7228820269200316e-05,
|
| 57 |
+
"loss": 0.3526,
|
| 58 |
+
"step": 350
|
| 59 |
+
},
|
| 60 |
+
{
|
| 61 |
+
"epoch": 0.4750593824228028,
|
| 62 |
+
"grad_norm": 4.217319011688232,
|
| 63 |
+
"learning_rate": 1.6832937450514647e-05,
|
| 64 |
+
"loss": 0.3525,
|
| 65 |
+
"step": 400
|
| 66 |
+
},
|
| 67 |
+
{
|
| 68 |
+
"epoch": 0.5344418052256532,
|
| 69 |
+
"grad_norm": 5.901436805725098,
|
| 70 |
+
"learning_rate": 1.643705463182898e-05,
|
| 71 |
+
"loss": 0.3351,
|
| 72 |
+
"step": 450
|
| 73 |
+
},
|
| 74 |
+
{
|
| 75 |
+
"epoch": 0.5938242280285035,
|
| 76 |
+
"grad_norm": 4.536603927612305,
|
| 77 |
+
"learning_rate": 1.604117181314331e-05,
|
| 78 |
+
"loss": 0.3287,
|
| 79 |
+
"step": 500
|
| 80 |
+
},
|
| 81 |
+
{
|
| 82 |
+
"epoch": 0.6532066508313539,
|
| 83 |
+
"grad_norm": 7.626627445220947,
|
| 84 |
+
"learning_rate": 1.564528899445764e-05,
|
| 85 |
+
"loss": 0.3254,
|
| 86 |
+
"step": 550
|
| 87 |
+
},
|
| 88 |
+
{
|
| 89 |
+
"epoch": 0.7125890736342043,
|
| 90 |
+
"grad_norm": 5.23235559463501,
|
| 91 |
+
"learning_rate": 1.5249406175771972e-05,
|
| 92 |
+
"loss": 0.3052,
|
| 93 |
+
"step": 600
|
| 94 |
+
},
|
| 95 |
+
{
|
| 96 |
+
"epoch": 0.7719714964370546,
|
| 97 |
+
"grad_norm": 5.8592448234558105,
|
| 98 |
+
"learning_rate": 1.4853523357086304e-05,
|
| 99 |
+
"loss": 0.3127,
|
| 100 |
+
"step": 650
|
| 101 |
+
},
|
| 102 |
+
{
|
| 103 |
+
"epoch": 0.831353919239905,
|
| 104 |
+
"grad_norm": 4.537547588348389,
|
| 105 |
+
"learning_rate": 1.4457640538400634e-05,
|
| 106 |
+
"loss": 0.3063,
|
| 107 |
+
"step": 700
|
| 108 |
+
},
|
| 109 |
+
{
|
| 110 |
+
"epoch": 0.8907363420427553,
|
| 111 |
+
"grad_norm": 5.8827362060546875,
|
| 112 |
+
"learning_rate": 1.4061757719714966e-05,
|
| 113 |
+
"loss": 0.3081,
|
| 114 |
+
"step": 750
|
| 115 |
+
},
|
| 116 |
+
{
|
| 117 |
+
"epoch": 0.9501187648456056,
|
| 118 |
+
"grad_norm": 6.283201694488525,
|
| 119 |
+
"learning_rate": 1.3665874901029297e-05,
|
| 120 |
+
"loss": 0.2945,
|
| 121 |
+
"step": 800
|
| 122 |
+
},
|
| 123 |
+
{
|
| 124 |
+
"epoch": 1.0,
|
| 125 |
+
"eval_accuracy": 0.875,
|
| 126 |
+
"eval_loss": 0.3256928324699402,
|
| 127 |
+
"eval_runtime": 2.4563,
|
| 128 |
+
"eval_samples_per_second": 355.009,
|
| 129 |
+
"eval_steps_per_second": 2.443,
|
| 130 |
+
"step": 842
|
| 131 |
+
},
|
| 132 |
+
{
|
| 133 |
+
"epoch": 1.009501187648456,
|
| 134 |
+
"grad_norm": 4.555658340454102,
|
| 135 |
+
"learning_rate": 1.3269992082343628e-05,
|
| 136 |
+
"loss": 0.2826,
|
| 137 |
+
"step": 850
|
| 138 |
+
},
|
| 139 |
+
{
|
| 140 |
+
"epoch": 1.0688836104513064,
|
| 141 |
+
"grad_norm": 4.976184368133545,
|
| 142 |
+
"learning_rate": 1.2874109263657959e-05,
|
| 143 |
+
"loss": 0.2493,
|
| 144 |
+
"step": 900
|
| 145 |
+
},
|
| 146 |
+
{
|
| 147 |
+
"epoch": 1.1282660332541568,
|
| 148 |
+
"grad_norm": 5.003899097442627,
|
| 149 |
+
"learning_rate": 1.247822644497229e-05,
|
| 150 |
+
"loss": 0.2344,
|
| 151 |
+
"step": 950
|
| 152 |
+
},
|
| 153 |
+
{
|
| 154 |
+
"epoch": 1.187648456057007,
|
| 155 |
+
"grad_norm": 4.125603675842285,
|
| 156 |
+
"learning_rate": 1.208234362628662e-05,
|
| 157 |
+
"loss": 0.251,
|
| 158 |
+
"step": 1000
|
| 159 |
+
},
|
| 160 |
+
{
|
| 161 |
+
"epoch": 1.2470308788598574,
|
| 162 |
+
"grad_norm": 6.115012168884277,
|
| 163 |
+
"learning_rate": 1.168646080760095e-05,
|
| 164 |
+
"loss": 0.2356,
|
| 165 |
+
"step": 1050
|
| 166 |
+
},
|
| 167 |
+
{
|
| 168 |
+
"epoch": 1.3064133016627077,
|
| 169 |
+
"grad_norm": 5.308842658996582,
|
| 170 |
+
"learning_rate": 1.1290577988915281e-05,
|
| 171 |
+
"loss": 0.2444,
|
| 172 |
+
"step": 1100
|
| 173 |
+
},
|
| 174 |
+
{
|
| 175 |
+
"epoch": 1.365795724465558,
|
| 176 |
+
"grad_norm": 7.800158500671387,
|
| 177 |
+
"learning_rate": 1.0894695170229613e-05,
|
| 178 |
+
"loss": 0.2403,
|
| 179 |
+
"step": 1150
|
| 180 |
+
},
|
| 181 |
+
{
|
| 182 |
+
"epoch": 1.4251781472684084,
|
| 183 |
+
"grad_norm": 6.025724411010742,
|
| 184 |
+
"learning_rate": 1.0498812351543943e-05,
|
| 185 |
+
"loss": 0.2465,
|
| 186 |
+
"step": 1200
|
| 187 |
+
},
|
| 188 |
+
{
|
| 189 |
+
"epoch": 1.484560570071259,
|
| 190 |
+
"grad_norm": 5.822863578796387,
|
| 191 |
+
"learning_rate": 1.0102929532858275e-05,
|
| 192 |
+
"loss": 0.225,
|
| 193 |
+
"step": 1250
|
| 194 |
+
},
|
| 195 |
+
{
|
| 196 |
+
"epoch": 1.5439429928741093,
|
| 197 |
+
"grad_norm": 4.0777740478515625,
|
| 198 |
+
"learning_rate": 9.707046714172605e-06,
|
| 199 |
+
"loss": 0.2371,
|
| 200 |
+
"step": 1300
|
| 201 |
+
},
|
| 202 |
+
{
|
| 203 |
+
"epoch": 1.6033254156769596,
|
| 204 |
+
"grad_norm": 6.322122097015381,
|
| 205 |
+
"learning_rate": 9.311163895486937e-06,
|
| 206 |
+
"loss": 0.2292,
|
| 207 |
+
"step": 1350
|
| 208 |
+
},
|
| 209 |
+
{
|
| 210 |
+
"epoch": 1.66270783847981,
|
| 211 |
+
"grad_norm": 5.80061149597168,
|
| 212 |
+
"learning_rate": 8.915281076801267e-06,
|
| 213 |
+
"loss": 0.2474,
|
| 214 |
+
"step": 1400
|
| 215 |
+
},
|
| 216 |
+
{
|
| 217 |
+
"epoch": 1.7220902612826603,
|
| 218 |
+
"grad_norm": 5.509008884429932,
|
| 219 |
+
"learning_rate": 8.5193982581156e-06,
|
| 220 |
+
"loss": 0.2213,
|
| 221 |
+
"step": 1450
|
| 222 |
+
},
|
| 223 |
+
{
|
| 224 |
+
"epoch": 1.7814726840855108,
|
| 225 |
+
"grad_norm": 6.122934341430664,
|
| 226 |
+
"learning_rate": 8.12351543942993e-06,
|
| 227 |
+
"loss": 0.2348,
|
| 228 |
+
"step": 1500
|
| 229 |
+
},
|
| 230 |
+
{
|
| 231 |
+
"epoch": 1.8408551068883612,
|
| 232 |
+
"grad_norm": 8.875970840454102,
|
| 233 |
+
"learning_rate": 7.72763262074426e-06,
|
| 234 |
+
"loss": 0.2094,
|
| 235 |
+
"step": 1550
|
| 236 |
+
},
|
| 237 |
+
{
|
| 238 |
+
"epoch": 1.9002375296912115,
|
| 239 |
+
"grad_norm": 5.162538528442383,
|
| 240 |
+
"learning_rate": 7.331749802058591e-06,
|
| 241 |
+
"loss": 0.2217,
|
| 242 |
+
"step": 1600
|
| 243 |
+
},
|
| 244 |
+
{
|
| 245 |
+
"epoch": 1.9596199524940618,
|
| 246 |
+
"grad_norm": 5.959411144256592,
|
| 247 |
+
"learning_rate": 6.935866983372922e-06,
|
| 248 |
+
"loss": 0.2125,
|
| 249 |
+
"step": 1650
|
| 250 |
+
},
|
| 251 |
+
{
|
| 252 |
+
"epoch": 2.0,
|
| 253 |
+
"eval_accuracy": 0.8727064220183486,
|
| 254 |
+
"eval_loss": 0.35895755887031555,
|
| 255 |
+
"eval_runtime": 2.4092,
|
| 256 |
+
"eval_samples_per_second": 361.953,
|
| 257 |
+
"eval_steps_per_second": 2.491,
|
| 258 |
+
"step": 1684
|
| 259 |
+
},
|
| 260 |
+
{
|
| 261 |
+
"epoch": 2.019002375296912,
|
| 262 |
+
"grad_norm": 3.9613893032073975,
|
| 263 |
+
"learning_rate": 6.539984164687253e-06,
|
| 264 |
+
"loss": 0.1972,
|
| 265 |
+
"step": 1700
|
| 266 |
+
},
|
| 267 |
+
{
|
| 268 |
+
"epoch": 2.0783847980997625,
|
| 269 |
+
"grad_norm": 6.392666816711426,
|
| 270 |
+
"learning_rate": 6.1441013460015846e-06,
|
| 271 |
+
"loss": 0.1719,
|
| 272 |
+
"step": 1750
|
| 273 |
+
},
|
| 274 |
+
{
|
| 275 |
+
"epoch": 2.137767220902613,
|
| 276 |
+
"grad_norm": 4.635701656341553,
|
| 277 |
+
"learning_rate": 5.748218527315916e-06,
|
| 278 |
+
"loss": 0.169,
|
| 279 |
+
"step": 1800
|
| 280 |
+
},
|
| 281 |
+
{
|
| 282 |
+
"epoch": 2.197149643705463,
|
| 283 |
+
"grad_norm": 7.464404582977295,
|
| 284 |
+
"learning_rate": 5.352335708630247e-06,
|
| 285 |
+
"loss": 0.1785,
|
| 286 |
+
"step": 1850
|
| 287 |
+
},
|
| 288 |
+
{
|
| 289 |
+
"epoch": 2.2565320665083135,
|
| 290 |
+
"grad_norm": 7.913965225219727,
|
| 291 |
+
"learning_rate": 4.956452889944577e-06,
|
| 292 |
+
"loss": 0.1691,
|
| 293 |
+
"step": 1900
|
| 294 |
+
},
|
| 295 |
+
{
|
| 296 |
+
"epoch": 2.315914489311164,
|
| 297 |
+
"grad_norm": 7.563793182373047,
|
| 298 |
+
"learning_rate": 4.560570071258908e-06,
|
| 299 |
+
"loss": 0.179,
|
| 300 |
+
"step": 1950
|
| 301 |
+
},
|
| 302 |
+
{
|
| 303 |
+
"epoch": 2.375296912114014,
|
| 304 |
+
"grad_norm": 3.7017734050750732,
|
| 305 |
+
"learning_rate": 4.164687252573238e-06,
|
| 306 |
+
"loss": 0.1705,
|
| 307 |
+
"step": 2000
|
| 308 |
+
},
|
| 309 |
+
{
|
| 310 |
+
"epoch": 2.4346793349168645,
|
| 311 |
+
"grad_norm": 5.2118353843688965,
|
| 312 |
+
"learning_rate": 3.7688044338875695e-06,
|
| 313 |
+
"loss": 0.1711,
|
| 314 |
+
"step": 2050
|
| 315 |
+
},
|
| 316 |
+
{
|
| 317 |
+
"epoch": 2.494061757719715,
|
| 318 |
+
"grad_norm": 5.063168048858643,
|
| 319 |
+
"learning_rate": 3.3729216152019006e-06,
|
| 320 |
+
"loss": 0.172,
|
| 321 |
+
"step": 2100
|
| 322 |
+
},
|
| 323 |
+
{
|
| 324 |
+
"epoch": 2.553444180522565,
|
| 325 |
+
"grad_norm": 6.718416690826416,
|
| 326 |
+
"learning_rate": 2.977038796516231e-06,
|
| 327 |
+
"loss": 0.1807,
|
| 328 |
+
"step": 2150
|
| 329 |
+
},
|
| 330 |
+
{
|
| 331 |
+
"epoch": 2.6128266033254155,
|
| 332 |
+
"grad_norm": 5.931078910827637,
|
| 333 |
+
"learning_rate": 2.581155977830562e-06,
|
| 334 |
+
"loss": 0.1697,
|
| 335 |
+
"step": 2200
|
| 336 |
+
},
|
| 337 |
+
{
|
| 338 |
+
"epoch": 2.6722090261282663,
|
| 339 |
+
"grad_norm": 8.941205978393555,
|
| 340 |
+
"learning_rate": 2.1852731591448932e-06,
|
| 341 |
+
"loss": 0.1552,
|
| 342 |
+
"step": 2250
|
| 343 |
+
},
|
| 344 |
+
{
|
| 345 |
+
"epoch": 2.731591448931116,
|
| 346 |
+
"grad_norm": 3.5849990844726562,
|
| 347 |
+
"learning_rate": 1.789390340459224e-06,
|
| 348 |
+
"loss": 0.1867,
|
| 349 |
+
"step": 2300
|
| 350 |
+
},
|
| 351 |
+
{
|
| 352 |
+
"epoch": 2.790973871733967,
|
| 353 |
+
"grad_norm": 8.088576316833496,
|
| 354 |
+
"learning_rate": 1.3935075217735553e-06,
|
| 355 |
+
"loss": 0.1705,
|
| 356 |
+
"step": 2350
|
| 357 |
+
},
|
| 358 |
+
{
|
| 359 |
+
"epoch": 2.850356294536817,
|
| 360 |
+
"grad_norm": 6.538223743438721,
|
| 361 |
+
"learning_rate": 9.976247030878861e-07,
|
| 362 |
+
"loss": 0.1721,
|
| 363 |
+
"step": 2400
|
| 364 |
+
},
|
| 365 |
+
{
|
| 366 |
+
"epoch": 2.9097387173396676,
|
| 367 |
+
"grad_norm": 6.613574981689453,
|
| 368 |
+
"learning_rate": 6.01741884402217e-07,
|
| 369 |
+
"loss": 0.1659,
|
| 370 |
+
"step": 2450
|
| 371 |
+
},
|
| 372 |
+
{
|
| 373 |
+
"epoch": 2.969121140142518,
|
| 374 |
+
"grad_norm": 2.8406691551208496,
|
| 375 |
+
"learning_rate": 2.0585906571654793e-07,
|
| 376 |
+
"loss": 0.1713,
|
| 377 |
+
"step": 2500
|
| 378 |
+
},
|
| 379 |
+
{
|
| 380 |
+
"epoch": 3.0,
|
| 381 |
+
"eval_accuracy": 0.8784403669724771,
|
| 382 |
+
"eval_loss": 0.3952513635158539,
|
| 383 |
+
"eval_runtime": 2.3672,
|
| 384 |
+
"eval_samples_per_second": 368.362,
|
| 385 |
+
"eval_steps_per_second": 2.535,
|
| 386 |
+
"step": 2526
|
| 387 |
+
},
|
| 388 |
+
{
|
| 389 |
+
"epoch": 3.0,
|
| 390 |
+
"step": 2526,
|
| 391 |
+
"total_flos": 6606837822286560.0,
|
| 392 |
+
"train_loss": 0.2572364091684397,
|
| 393 |
+
"train_runtime": 266.5223,
|
| 394 |
+
"train_samples_per_second": 758.087,
|
| 395 |
+
"train_steps_per_second": 9.478
|
| 396 |
+
}
|
| 397 |
+
],
|
| 398 |
+
"logging_steps": 50,
|
| 399 |
+
"max_steps": 2526,
|
| 400 |
+
"num_input_tokens_seen": 0,
|
| 401 |
+
"num_train_epochs": 3,
|
| 402 |
+
"save_steps": 500,
|
| 403 |
+
"stateful_callbacks": {
|
| 404 |
+
"TrainerControl": {
|
| 405 |
+
"args": {
|
| 406 |
+
"should_epoch_stop": false,
|
| 407 |
+
"should_evaluate": false,
|
| 408 |
+
"should_log": false,
|
| 409 |
+
"should_save": false,
|
| 410 |
+
"should_training_stop": false
|
| 411 |
+
},
|
| 412 |
+
"attributes": {}
|
| 413 |
+
}
|
| 414 |
+
},
|
| 415 |
+
"total_flos": 6606837822286560.0,
|
| 416 |
+
"train_batch_size": 80,
|
| 417 |
+
"trial_name": null,
|
| 418 |
+
"trial_params": null
|
| 419 |
+
}
|
mbert_paper_metrics/result_ablation_mbert_paper/sst2/mBERT/hf_sequence_classifier/seed_44/all_results.json
ADDED
|
@@ -0,0 +1,15 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 3.0,
|
| 3 |
+
"eval_accuracy": 0.8887614678899083,
|
| 4 |
+
"eval_loss": 0.3691374659538269,
|
| 5 |
+
"eval_runtime": 2.159,
|
| 6 |
+
"eval_samples": 872,
|
| 7 |
+
"eval_samples_per_second": 403.898,
|
| 8 |
+
"eval_steps_per_second": 2.779,
|
| 9 |
+
"total_flos": 6603598781306220.0,
|
| 10 |
+
"train_loss": 0.2579068086115217,
|
| 11 |
+
"train_runtime": 266.0721,
|
| 12 |
+
"train_samples": 67349,
|
| 13 |
+
"train_samples_per_second": 759.369,
|
| 14 |
+
"train_steps_per_second": 9.494
|
| 15 |
+
}
|
mbert_paper_metrics/result_ablation_mbert_paper/sst2/mBERT/hf_sequence_classifier/seed_44/eval_results.json
ADDED
|
@@ -0,0 +1,9 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 3.0,
|
| 3 |
+
"eval_accuracy": 0.8887614678899083,
|
| 4 |
+
"eval_loss": 0.3691374659538269,
|
| 5 |
+
"eval_runtime": 2.159,
|
| 6 |
+
"eval_samples": 872,
|
| 7 |
+
"eval_samples_per_second": 403.898,
|
| 8 |
+
"eval_steps_per_second": 2.779
|
| 9 |
+
}
|
mbert_paper_metrics/result_ablation_mbert_paper/sst2/mBERT/hf_sequence_classifier/seed_44/run_command.txt
ADDED
|
@@ -0,0 +1 @@
|
|
|
|
|
|
|
| 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 sst2 --max_seq_length 256 --do_train --do_eval --evaluation_strategy epoch --save_strategy no --logging_strategy steps --logging_steps 50 --report_to none --num_train_epochs 3 --learning_rate 2e-05 --per_device_train_batch_size 80 --per_device_eval_batch_size 168 --gradient_accumulation_steps 1 --dataloader_num_workers 2 --seed 44 --output_dir result_ablation_mbert_paper/sst2/mBERT/hf_sequence_classifier/seed_44 --overwrite_output_dir --trust_remote_code true --bf16 --tf32 true --skip_model_save
|
mbert_paper_metrics/result_ablation_mbert_paper/sst2/mBERT/hf_sequence_classifier/seed_44/train_results.json
ADDED
|
@@ -0,0 +1,9 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 3.0,
|
| 3 |
+
"total_flos": 6603598781306220.0,
|
| 4 |
+
"train_loss": 0.2579068086115217,
|
| 5 |
+
"train_runtime": 266.0721,
|
| 6 |
+
"train_samples": 67349,
|
| 7 |
+
"train_samples_per_second": 759.369,
|
| 8 |
+
"train_steps_per_second": 9.494
|
| 9 |
+
}
|
mbert_paper_metrics/result_ablation_mbert_paper/sst2/mBERT/hf_sequence_classifier/seed_44/trainer_state.json
ADDED
|
@@ -0,0 +1,419 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_metric": null,
|
| 3 |
+
"best_model_checkpoint": null,
|
| 4 |
+
"epoch": 3.0,
|
| 5 |
+
"eval_steps": 500,
|
| 6 |
+
"global_step": 2526,
|
| 7 |
+
"is_hyper_param_search": false,
|
| 8 |
+
"is_local_process_zero": true,
|
| 9 |
+
"is_world_process_zero": true,
|
| 10 |
+
"log_history": [
|
| 11 |
+
{
|
| 12 |
+
"epoch": 0.05938242280285035,
|
| 13 |
+
"grad_norm": 4.057610034942627,
|
| 14 |
+
"learning_rate": 1.9604117181314332e-05,
|
| 15 |
+
"loss": 0.5993,
|
| 16 |
+
"step": 50
|
| 17 |
+
},
|
| 18 |
+
{
|
| 19 |
+
"epoch": 0.1187648456057007,
|
| 20 |
+
"grad_norm": 4.086034774780273,
|
| 21 |
+
"learning_rate": 1.9208234362628665e-05,
|
| 22 |
+
"loss": 0.4779,
|
| 23 |
+
"step": 100
|
| 24 |
+
},
|
| 25 |
+
{
|
| 26 |
+
"epoch": 0.17814726840855108,
|
| 27 |
+
"grad_norm": 3.9259910583496094,
|
| 28 |
+
"learning_rate": 1.8812351543942996e-05,
|
| 29 |
+
"loss": 0.4284,
|
| 30 |
+
"step": 150
|
| 31 |
+
},
|
| 32 |
+
{
|
| 33 |
+
"epoch": 0.2375296912114014,
|
| 34 |
+
"grad_norm": 4.98960542678833,
|
| 35 |
+
"learning_rate": 1.8416468725257326e-05,
|
| 36 |
+
"loss": 0.4045,
|
| 37 |
+
"step": 200
|
| 38 |
+
},
|
| 39 |
+
{
|
| 40 |
+
"epoch": 0.29691211401425177,
|
| 41 |
+
"grad_norm": 3.7215912342071533,
|
| 42 |
+
"learning_rate": 1.8020585906571656e-05,
|
| 43 |
+
"loss": 0.3834,
|
| 44 |
+
"step": 250
|
| 45 |
+
},
|
| 46 |
+
{
|
| 47 |
+
"epoch": 0.35629453681710216,
|
| 48 |
+
"grad_norm": 5.476784706115723,
|
| 49 |
+
"learning_rate": 1.762470308788599e-05,
|
| 50 |
+
"loss": 0.3839,
|
| 51 |
+
"step": 300
|
| 52 |
+
},
|
| 53 |
+
{
|
| 54 |
+
"epoch": 0.4156769596199525,
|
| 55 |
+
"grad_norm": 4.6505842208862305,
|
| 56 |
+
"learning_rate": 1.7228820269200316e-05,
|
| 57 |
+
"loss": 0.3693,
|
| 58 |
+
"step": 350
|
| 59 |
+
},
|
| 60 |
+
{
|
| 61 |
+
"epoch": 0.4750593824228028,
|
| 62 |
+
"grad_norm": 4.8869805335998535,
|
| 63 |
+
"learning_rate": 1.6832937450514647e-05,
|
| 64 |
+
"loss": 0.3518,
|
| 65 |
+
"step": 400
|
| 66 |
+
},
|
| 67 |
+
{
|
| 68 |
+
"epoch": 0.5344418052256532,
|
| 69 |
+
"grad_norm": 4.835771083831787,
|
| 70 |
+
"learning_rate": 1.643705463182898e-05,
|
| 71 |
+
"loss": 0.3338,
|
| 72 |
+
"step": 450
|
| 73 |
+
},
|
| 74 |
+
{
|
| 75 |
+
"epoch": 0.5938242280285035,
|
| 76 |
+
"grad_norm": 6.663527965545654,
|
| 77 |
+
"learning_rate": 1.604117181314331e-05,
|
| 78 |
+
"loss": 0.3407,
|
| 79 |
+
"step": 500
|
| 80 |
+
},
|
| 81 |
+
{
|
| 82 |
+
"epoch": 0.6532066508313539,
|
| 83 |
+
"grad_norm": 4.45944881439209,
|
| 84 |
+
"learning_rate": 1.564528899445764e-05,
|
| 85 |
+
"loss": 0.3289,
|
| 86 |
+
"step": 550
|
| 87 |
+
},
|
| 88 |
+
{
|
| 89 |
+
"epoch": 0.7125890736342043,
|
| 90 |
+
"grad_norm": 6.935768127441406,
|
| 91 |
+
"learning_rate": 1.5249406175771972e-05,
|
| 92 |
+
"loss": 0.3165,
|
| 93 |
+
"step": 600
|
| 94 |
+
},
|
| 95 |
+
{
|
| 96 |
+
"epoch": 0.7719714964370546,
|
| 97 |
+
"grad_norm": 4.2998738288879395,
|
| 98 |
+
"learning_rate": 1.4853523357086304e-05,
|
| 99 |
+
"loss": 0.3169,
|
| 100 |
+
"step": 650
|
| 101 |
+
},
|
| 102 |
+
{
|
| 103 |
+
"epoch": 0.831353919239905,
|
| 104 |
+
"grad_norm": 5.204697132110596,
|
| 105 |
+
"learning_rate": 1.4457640538400634e-05,
|
| 106 |
+
"loss": 0.3067,
|
| 107 |
+
"step": 700
|
| 108 |
+
},
|
| 109 |
+
{
|
| 110 |
+
"epoch": 0.8907363420427553,
|
| 111 |
+
"grad_norm": 5.685110569000244,
|
| 112 |
+
"learning_rate": 1.4061757719714966e-05,
|
| 113 |
+
"loss": 0.3002,
|
| 114 |
+
"step": 750
|
| 115 |
+
},
|
| 116 |
+
{
|
| 117 |
+
"epoch": 0.9501187648456056,
|
| 118 |
+
"grad_norm": 4.944071292877197,
|
| 119 |
+
"learning_rate": 1.3665874901029297e-05,
|
| 120 |
+
"loss": 0.3055,
|
| 121 |
+
"step": 800
|
| 122 |
+
},
|
| 123 |
+
{
|
| 124 |
+
"epoch": 1.0,
|
| 125 |
+
"eval_accuracy": 0.8589449541284404,
|
| 126 |
+
"eval_loss": 0.342286080121994,
|
| 127 |
+
"eval_runtime": 2.4486,
|
| 128 |
+
"eval_samples_per_second": 356.119,
|
| 129 |
+
"eval_steps_per_second": 2.45,
|
| 130 |
+
"step": 842
|
| 131 |
+
},
|
| 132 |
+
{
|
| 133 |
+
"epoch": 1.009501187648456,
|
| 134 |
+
"grad_norm": 7.254140853881836,
|
| 135 |
+
"learning_rate": 1.3269992082343628e-05,
|
| 136 |
+
"loss": 0.2841,
|
| 137 |
+
"step": 850
|
| 138 |
+
},
|
| 139 |
+
{
|
| 140 |
+
"epoch": 1.0688836104513064,
|
| 141 |
+
"grad_norm": 6.600419521331787,
|
| 142 |
+
"learning_rate": 1.2874109263657959e-05,
|
| 143 |
+
"loss": 0.2475,
|
| 144 |
+
"step": 900
|
| 145 |
+
},
|
| 146 |
+
{
|
| 147 |
+
"epoch": 1.1282660332541568,
|
| 148 |
+
"grad_norm": 5.566004753112793,
|
| 149 |
+
"learning_rate": 1.247822644497229e-05,
|
| 150 |
+
"loss": 0.2265,
|
| 151 |
+
"step": 950
|
| 152 |
+
},
|
| 153 |
+
{
|
| 154 |
+
"epoch": 1.187648456057007,
|
| 155 |
+
"grad_norm": 4.097609996795654,
|
| 156 |
+
"learning_rate": 1.208234362628662e-05,
|
| 157 |
+
"loss": 0.2457,
|
| 158 |
+
"step": 1000
|
| 159 |
+
},
|
| 160 |
+
{
|
| 161 |
+
"epoch": 1.2470308788598574,
|
| 162 |
+
"grad_norm": 3.9480111598968506,
|
| 163 |
+
"learning_rate": 1.168646080760095e-05,
|
| 164 |
+
"loss": 0.2425,
|
| 165 |
+
"step": 1050
|
| 166 |
+
},
|
| 167 |
+
{
|
| 168 |
+
"epoch": 1.3064133016627077,
|
| 169 |
+
"grad_norm": 4.7737812995910645,
|
| 170 |
+
"learning_rate": 1.1290577988915281e-05,
|
| 171 |
+
"loss": 0.2413,
|
| 172 |
+
"step": 1100
|
| 173 |
+
},
|
| 174 |
+
{
|
| 175 |
+
"epoch": 1.365795724465558,
|
| 176 |
+
"grad_norm": 5.022617340087891,
|
| 177 |
+
"learning_rate": 1.0894695170229613e-05,
|
| 178 |
+
"loss": 0.2456,
|
| 179 |
+
"step": 1150
|
| 180 |
+
},
|
| 181 |
+
{
|
| 182 |
+
"epoch": 1.4251781472684084,
|
| 183 |
+
"grad_norm": 5.507868766784668,
|
| 184 |
+
"learning_rate": 1.0498812351543943e-05,
|
| 185 |
+
"loss": 0.2423,
|
| 186 |
+
"step": 1200
|
| 187 |
+
},
|
| 188 |
+
{
|
| 189 |
+
"epoch": 1.484560570071259,
|
| 190 |
+
"grad_norm": 4.600748062133789,
|
| 191 |
+
"learning_rate": 1.0102929532858275e-05,
|
| 192 |
+
"loss": 0.2316,
|
| 193 |
+
"step": 1250
|
| 194 |
+
},
|
| 195 |
+
{
|
| 196 |
+
"epoch": 1.5439429928741093,
|
| 197 |
+
"grad_norm": 5.917996883392334,
|
| 198 |
+
"learning_rate": 9.707046714172605e-06,
|
| 199 |
+
"loss": 0.2351,
|
| 200 |
+
"step": 1300
|
| 201 |
+
},
|
| 202 |
+
{
|
| 203 |
+
"epoch": 1.6033254156769596,
|
| 204 |
+
"grad_norm": 5.940685749053955,
|
| 205 |
+
"learning_rate": 9.311163895486937e-06,
|
| 206 |
+
"loss": 0.2258,
|
| 207 |
+
"step": 1350
|
| 208 |
+
},
|
| 209 |
+
{
|
| 210 |
+
"epoch": 1.66270783847981,
|
| 211 |
+
"grad_norm": 4.993325710296631,
|
| 212 |
+
"learning_rate": 8.915281076801267e-06,
|
| 213 |
+
"loss": 0.2191,
|
| 214 |
+
"step": 1400
|
| 215 |
+
},
|
| 216 |
+
{
|
| 217 |
+
"epoch": 1.7220902612826603,
|
| 218 |
+
"grad_norm": 5.060179710388184,
|
| 219 |
+
"learning_rate": 8.5193982581156e-06,
|
| 220 |
+
"loss": 0.2298,
|
| 221 |
+
"step": 1450
|
| 222 |
+
},
|
| 223 |
+
{
|
| 224 |
+
"epoch": 1.7814726840855108,
|
| 225 |
+
"grad_norm": 4.853836536407471,
|
| 226 |
+
"learning_rate": 8.12351543942993e-06,
|
| 227 |
+
"loss": 0.2269,
|
| 228 |
+
"step": 1500
|
| 229 |
+
},
|
| 230 |
+
{
|
| 231 |
+
"epoch": 1.8408551068883612,
|
| 232 |
+
"grad_norm": 4.663071632385254,
|
| 233 |
+
"learning_rate": 7.72763262074426e-06,
|
| 234 |
+
"loss": 0.2245,
|
| 235 |
+
"step": 1550
|
| 236 |
+
},
|
| 237 |
+
{
|
| 238 |
+
"epoch": 1.9002375296912115,
|
| 239 |
+
"grad_norm": 6.5991973876953125,
|
| 240 |
+
"learning_rate": 7.331749802058591e-06,
|
| 241 |
+
"loss": 0.2218,
|
| 242 |
+
"step": 1600
|
| 243 |
+
},
|
| 244 |
+
{
|
| 245 |
+
"epoch": 1.9596199524940618,
|
| 246 |
+
"grad_norm": 3.9060027599334717,
|
| 247 |
+
"learning_rate": 6.935866983372922e-06,
|
| 248 |
+
"loss": 0.2232,
|
| 249 |
+
"step": 1650
|
| 250 |
+
},
|
| 251 |
+
{
|
| 252 |
+
"epoch": 2.0,
|
| 253 |
+
"eval_accuracy": 0.8727064220183486,
|
| 254 |
+
"eval_loss": 0.35681799054145813,
|
| 255 |
+
"eval_runtime": 2.3479,
|
| 256 |
+
"eval_samples_per_second": 371.392,
|
| 257 |
+
"eval_steps_per_second": 2.555,
|
| 258 |
+
"step": 1684
|
| 259 |
+
},
|
| 260 |
+
{
|
| 261 |
+
"epoch": 2.019002375296912,
|
| 262 |
+
"grad_norm": 3.4162986278533936,
|
| 263 |
+
"learning_rate": 6.539984164687253e-06,
|
| 264 |
+
"loss": 0.2028,
|
| 265 |
+
"step": 1700
|
| 266 |
+
},
|
| 267 |
+
{
|
| 268 |
+
"epoch": 2.0783847980997625,
|
| 269 |
+
"grad_norm": 7.023327350616455,
|
| 270 |
+
"learning_rate": 6.1441013460015846e-06,
|
| 271 |
+
"loss": 0.19,
|
| 272 |
+
"step": 1750
|
| 273 |
+
},
|
| 274 |
+
{
|
| 275 |
+
"epoch": 2.137767220902613,
|
| 276 |
+
"grad_norm": 8.123522758483887,
|
| 277 |
+
"learning_rate": 5.748218527315916e-06,
|
| 278 |
+
"loss": 0.1687,
|
| 279 |
+
"step": 1800
|
| 280 |
+
},
|
| 281 |
+
{
|
| 282 |
+
"epoch": 2.197149643705463,
|
| 283 |
+
"grad_norm": 9.088725090026855,
|
| 284 |
+
"learning_rate": 5.352335708630247e-06,
|
| 285 |
+
"loss": 0.1891,
|
| 286 |
+
"step": 1850
|
| 287 |
+
},
|
| 288 |
+
{
|
| 289 |
+
"epoch": 2.2565320665083135,
|
| 290 |
+
"grad_norm": 7.173944473266602,
|
| 291 |
+
"learning_rate": 4.956452889944577e-06,
|
| 292 |
+
"loss": 0.1667,
|
| 293 |
+
"step": 1900
|
| 294 |
+
},
|
| 295 |
+
{
|
| 296 |
+
"epoch": 2.315914489311164,
|
| 297 |
+
"grad_norm": 7.316494464874268,
|
| 298 |
+
"learning_rate": 4.560570071258908e-06,
|
| 299 |
+
"loss": 0.171,
|
| 300 |
+
"step": 1950
|
| 301 |
+
},
|
| 302 |
+
{
|
| 303 |
+
"epoch": 2.375296912114014,
|
| 304 |
+
"grad_norm": 3.819225788116455,
|
| 305 |
+
"learning_rate": 4.164687252573238e-06,
|
| 306 |
+
"loss": 0.1687,
|
| 307 |
+
"step": 2000
|
| 308 |
+
},
|
| 309 |
+
{
|
| 310 |
+
"epoch": 2.4346793349168645,
|
| 311 |
+
"grad_norm": 4.682591438293457,
|
| 312 |
+
"learning_rate": 3.7688044338875695e-06,
|
| 313 |
+
"loss": 0.169,
|
| 314 |
+
"step": 2050
|
| 315 |
+
},
|
| 316 |
+
{
|
| 317 |
+
"epoch": 2.494061757719715,
|
| 318 |
+
"grad_norm": 5.729262828826904,
|
| 319 |
+
"learning_rate": 3.3729216152019006e-06,
|
| 320 |
+
"loss": 0.1775,
|
| 321 |
+
"step": 2100
|
| 322 |
+
},
|
| 323 |
+
{
|
| 324 |
+
"epoch": 2.553444180522565,
|
| 325 |
+
"grad_norm": 8.35857105255127,
|
| 326 |
+
"learning_rate": 2.977038796516231e-06,
|
| 327 |
+
"loss": 0.1643,
|
| 328 |
+
"step": 2150
|
| 329 |
+
},
|
| 330 |
+
{
|
| 331 |
+
"epoch": 2.6128266033254155,
|
| 332 |
+
"grad_norm": 6.792198181152344,
|
| 333 |
+
"learning_rate": 2.581155977830562e-06,
|
| 334 |
+
"loss": 0.1692,
|
| 335 |
+
"step": 2200
|
| 336 |
+
},
|
| 337 |
+
{
|
| 338 |
+
"epoch": 2.6722090261282663,
|
| 339 |
+
"grad_norm": 4.841739177703857,
|
| 340 |
+
"learning_rate": 2.1852731591448932e-06,
|
| 341 |
+
"loss": 0.1717,
|
| 342 |
+
"step": 2250
|
| 343 |
+
},
|
| 344 |
+
{
|
| 345 |
+
"epoch": 2.731591448931116,
|
| 346 |
+
"grad_norm": 4.04350471496582,
|
| 347 |
+
"learning_rate": 1.789390340459224e-06,
|
| 348 |
+
"loss": 0.1874,
|
| 349 |
+
"step": 2300
|
| 350 |
+
},
|
| 351 |
+
{
|
| 352 |
+
"epoch": 2.790973871733967,
|
| 353 |
+
"grad_norm": 4.533483505249023,
|
| 354 |
+
"learning_rate": 1.3935075217735553e-06,
|
| 355 |
+
"loss": 0.1677,
|
| 356 |
+
"step": 2350
|
| 357 |
+
},
|
| 358 |
+
{
|
| 359 |
+
"epoch": 2.850356294536817,
|
| 360 |
+
"grad_norm": 6.155765533447266,
|
| 361 |
+
"learning_rate": 9.976247030878861e-07,
|
| 362 |
+
"loss": 0.1721,
|
| 363 |
+
"step": 2400
|
| 364 |
+
},
|
| 365 |
+
{
|
| 366 |
+
"epoch": 2.9097387173396676,
|
| 367 |
+
"grad_norm": 4.566720962524414,
|
| 368 |
+
"learning_rate": 6.01741884402217e-07,
|
| 369 |
+
"loss": 0.1744,
|
| 370 |
+
"step": 2450
|
| 371 |
+
},
|
| 372 |
+
{
|
| 373 |
+
"epoch": 2.969121140142518,
|
| 374 |
+
"grad_norm": 7.027346611022949,
|
| 375 |
+
"learning_rate": 2.0585906571654793e-07,
|
| 376 |
+
"loss": 0.1672,
|
| 377 |
+
"step": 2500
|
| 378 |
+
},
|
| 379 |
+
{
|
| 380 |
+
"epoch": 3.0,
|
| 381 |
+
"eval_accuracy": 0.8887614678899083,
|
| 382 |
+
"eval_loss": 0.3691374659538269,
|
| 383 |
+
"eval_runtime": 2.2214,
|
| 384 |
+
"eval_samples_per_second": 392.54,
|
| 385 |
+
"eval_steps_per_second": 2.701,
|
| 386 |
+
"step": 2526
|
| 387 |
+
},
|
| 388 |
+
{
|
| 389 |
+
"epoch": 3.0,
|
| 390 |
+
"step": 2526,
|
| 391 |
+
"total_flos": 6603598781306220.0,
|
| 392 |
+
"train_loss": 0.2579068086115217,
|
| 393 |
+
"train_runtime": 266.0721,
|
| 394 |
+
"train_samples_per_second": 759.369,
|
| 395 |
+
"train_steps_per_second": 9.494
|
| 396 |
+
}
|
| 397 |
+
],
|
| 398 |
+
"logging_steps": 50,
|
| 399 |
+
"max_steps": 2526,
|
| 400 |
+
"num_input_tokens_seen": 0,
|
| 401 |
+
"num_train_epochs": 3,
|
| 402 |
+
"save_steps": 500,
|
| 403 |
+
"stateful_callbacks": {
|
| 404 |
+
"TrainerControl": {
|
| 405 |
+
"args": {
|
| 406 |
+
"should_epoch_stop": false,
|
| 407 |
+
"should_evaluate": false,
|
| 408 |
+
"should_log": false,
|
| 409 |
+
"should_save": false,
|
| 410 |
+
"should_training_stop": false
|
| 411 |
+
},
|
| 412 |
+
"attributes": {}
|
| 413 |
+
}
|
| 414 |
+
},
|
| 415 |
+
"total_flos": 6603598781306220.0,
|
| 416 |
+
"train_batch_size": 80,
|
| 417 |
+
"trial_name": null,
|
| 418 |
+
"trial_params": null
|
| 419 |
+
}
|
mbert_paper_metrics/result_ablation_mbert_paper/sst2/mBERT/max/seed_42/all_results.json
ADDED
|
@@ -0,0 +1,15 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 3.0,
|
| 3 |
+
"eval_accuracy": 0.8704128440366973,
|
| 4 |
+
"eval_loss": 0.39564406871795654,
|
| 5 |
+
"eval_runtime": 2.0024,
|
| 6 |
+
"eval_samples": 872,
|
| 7 |
+
"eval_samples_per_second": 435.488,
|
| 8 |
+
"eval_steps_per_second": 2.996,
|
| 9 |
+
"total_flos": 6631676508966912.0,
|
| 10 |
+
"train_loss": 0.25985605001260814,
|
| 11 |
+
"train_runtime": 258.2146,
|
| 12 |
+
"train_samples": 67349,
|
| 13 |
+
"train_samples_per_second": 782.477,
|
| 14 |
+
"train_steps_per_second": 9.783
|
| 15 |
+
}
|
mbert_paper_metrics/result_ablation_mbert_paper/sst2/mBERT/max/seed_42/eval_results.json
ADDED
|
@@ -0,0 +1,9 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 3.0,
|
| 3 |
+
"eval_accuracy": 0.8704128440366973,
|
| 4 |
+
"eval_loss": 0.39564406871795654,
|
| 5 |
+
"eval_runtime": 2.0024,
|
| 6 |
+
"eval_samples": 872,
|
| 7 |
+
"eval_samples_per_second": 435.488,
|
| 8 |
+
"eval_steps_per_second": 2.996
|
| 9 |
+
}
|
mbert_paper_metrics/result_ablation_mbert_paper/sst2/mBERT/max/seed_42/run_command.txt
ADDED
|
@@ -0,0 +1 @@
|
|
|
|
|
|
|
| 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 sst2 --max_seq_length 256 --do_train --do_eval --evaluation_strategy epoch --save_strategy no --logging_strategy steps --logging_steps 50 --report_to none --num_train_epochs 3 --learning_rate 2e-05 --per_device_train_batch_size 80 --per_device_eval_batch_size 168 --gradient_accumulation_steps 1 --dataloader_num_workers 2 --seed 42 --output_dir result_ablation_mbert_paper/sst2/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 8 --head_dim 32 --classifier_hidden_dim 512 --head_dropout 0.05
|
mbert_paper_metrics/result_ablation_mbert_paper/sst2/mBERT/max/seed_42/train_results.json
ADDED
|
@@ -0,0 +1,9 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 3.0,
|
| 3 |
+
"total_flos": 6631676508966912.0,
|
| 4 |
+
"train_loss": 0.25985605001260814,
|
| 5 |
+
"train_runtime": 258.2146,
|
| 6 |
+
"train_samples": 67349,
|
| 7 |
+
"train_samples_per_second": 782.477,
|
| 8 |
+
"train_steps_per_second": 9.783
|
| 9 |
+
}
|
mbert_paper_metrics/result_ablation_mbert_paper/sst2/mBERT/max/seed_42/trainer_state.json
ADDED
|
@@ -0,0 +1,419 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_metric": null,
|
| 3 |
+
"best_model_checkpoint": null,
|
| 4 |
+
"epoch": 3.0,
|
| 5 |
+
"eval_steps": 500,
|
| 6 |
+
"global_step": 2526,
|
| 7 |
+
"is_hyper_param_search": false,
|
| 8 |
+
"is_local_process_zero": true,
|
| 9 |
+
"is_world_process_zero": true,
|
| 10 |
+
"log_history": [
|
| 11 |
+
{
|
| 12 |
+
"epoch": 0.05938242280285035,
|
| 13 |
+
"grad_norm": 4.36276388168335,
|
| 14 |
+
"learning_rate": 1.9604117181314332e-05,
|
| 15 |
+
"loss": 0.605,
|
| 16 |
+
"step": 50
|
| 17 |
+
},
|
| 18 |
+
{
|
| 19 |
+
"epoch": 0.1187648456057007,
|
| 20 |
+
"grad_norm": 5.29181432723999,
|
| 21 |
+
"learning_rate": 1.9208234362628665e-05,
|
| 22 |
+
"loss": 0.4782,
|
| 23 |
+
"step": 100
|
| 24 |
+
},
|
| 25 |
+
{
|
| 26 |
+
"epoch": 0.17814726840855108,
|
| 27 |
+
"grad_norm": 7.856871604919434,
|
| 28 |
+
"learning_rate": 1.8812351543942996e-05,
|
| 29 |
+
"loss": 0.4275,
|
| 30 |
+
"step": 150
|
| 31 |
+
},
|
| 32 |
+
{
|
| 33 |
+
"epoch": 0.2375296912114014,
|
| 34 |
+
"grad_norm": 5.300786972045898,
|
| 35 |
+
"learning_rate": 1.8416468725257326e-05,
|
| 36 |
+
"loss": 0.4238,
|
| 37 |
+
"step": 200
|
| 38 |
+
},
|
| 39 |
+
{
|
| 40 |
+
"epoch": 0.29691211401425177,
|
| 41 |
+
"grad_norm": 6.970973014831543,
|
| 42 |
+
"learning_rate": 1.8020585906571656e-05,
|
| 43 |
+
"loss": 0.3964,
|
| 44 |
+
"step": 250
|
| 45 |
+
},
|
| 46 |
+
{
|
| 47 |
+
"epoch": 0.35629453681710216,
|
| 48 |
+
"grad_norm": 5.289291858673096,
|
| 49 |
+
"learning_rate": 1.762470308788599e-05,
|
| 50 |
+
"loss": 0.3815,
|
| 51 |
+
"step": 300
|
| 52 |
+
},
|
| 53 |
+
{
|
| 54 |
+
"epoch": 0.4156769596199525,
|
| 55 |
+
"grad_norm": 4.635684013366699,
|
| 56 |
+
"learning_rate": 1.7228820269200316e-05,
|
| 57 |
+
"loss": 0.346,
|
| 58 |
+
"step": 350
|
| 59 |
+
},
|
| 60 |
+
{
|
| 61 |
+
"epoch": 0.4750593824228028,
|
| 62 |
+
"grad_norm": 4.960316181182861,
|
| 63 |
+
"learning_rate": 1.6832937450514647e-05,
|
| 64 |
+
"loss": 0.3455,
|
| 65 |
+
"step": 400
|
| 66 |
+
},
|
| 67 |
+
{
|
| 68 |
+
"epoch": 0.5344418052256532,
|
| 69 |
+
"grad_norm": 5.120222568511963,
|
| 70 |
+
"learning_rate": 1.643705463182898e-05,
|
| 71 |
+
"loss": 0.3435,
|
| 72 |
+
"step": 450
|
| 73 |
+
},
|
| 74 |
+
{
|
| 75 |
+
"epoch": 0.5938242280285035,
|
| 76 |
+
"grad_norm": 4.728334426879883,
|
| 77 |
+
"learning_rate": 1.604117181314331e-05,
|
| 78 |
+
"loss": 0.3298,
|
| 79 |
+
"step": 500
|
| 80 |
+
},
|
| 81 |
+
{
|
| 82 |
+
"epoch": 0.6532066508313539,
|
| 83 |
+
"grad_norm": 3.9678382873535156,
|
| 84 |
+
"learning_rate": 1.564528899445764e-05,
|
| 85 |
+
"loss": 0.3348,
|
| 86 |
+
"step": 550
|
| 87 |
+
},
|
| 88 |
+
{
|
| 89 |
+
"epoch": 0.7125890736342043,
|
| 90 |
+
"grad_norm": 4.005518436431885,
|
| 91 |
+
"learning_rate": 1.5249406175771972e-05,
|
| 92 |
+
"loss": 0.3328,
|
| 93 |
+
"step": 600
|
| 94 |
+
},
|
| 95 |
+
{
|
| 96 |
+
"epoch": 0.7719714964370546,
|
| 97 |
+
"grad_norm": 4.141592979431152,
|
| 98 |
+
"learning_rate": 1.4853523357086304e-05,
|
| 99 |
+
"loss": 0.315,
|
| 100 |
+
"step": 650
|
| 101 |
+
},
|
| 102 |
+
{
|
| 103 |
+
"epoch": 0.831353919239905,
|
| 104 |
+
"grad_norm": 3.9195730686187744,
|
| 105 |
+
"learning_rate": 1.4457640538400634e-05,
|
| 106 |
+
"loss": 0.3188,
|
| 107 |
+
"step": 700
|
| 108 |
+
},
|
| 109 |
+
{
|
| 110 |
+
"epoch": 0.8907363420427553,
|
| 111 |
+
"grad_norm": 4.1250319480896,
|
| 112 |
+
"learning_rate": 1.4061757719714966e-05,
|
| 113 |
+
"loss": 0.31,
|
| 114 |
+
"step": 750
|
| 115 |
+
},
|
| 116 |
+
{
|
| 117 |
+
"epoch": 0.9501187648456056,
|
| 118 |
+
"grad_norm": 4.071584701538086,
|
| 119 |
+
"learning_rate": 1.3665874901029297e-05,
|
| 120 |
+
"loss": 0.2916,
|
| 121 |
+
"step": 800
|
| 122 |
+
},
|
| 123 |
+
{
|
| 124 |
+
"epoch": 1.0,
|
| 125 |
+
"eval_accuracy": 0.8555045871559633,
|
| 126 |
+
"eval_loss": 0.3355642259120941,
|
| 127 |
+
"eval_runtime": 2.0447,
|
| 128 |
+
"eval_samples_per_second": 426.472,
|
| 129 |
+
"eval_steps_per_second": 2.934,
|
| 130 |
+
"step": 842
|
| 131 |
+
},
|
| 132 |
+
{
|
| 133 |
+
"epoch": 1.009501187648456,
|
| 134 |
+
"grad_norm": 3.5726358890533447,
|
| 135 |
+
"learning_rate": 1.3269992082343628e-05,
|
| 136 |
+
"loss": 0.3008,
|
| 137 |
+
"step": 850
|
| 138 |
+
},
|
| 139 |
+
{
|
| 140 |
+
"epoch": 1.0688836104513064,
|
| 141 |
+
"grad_norm": 4.480276107788086,
|
| 142 |
+
"learning_rate": 1.2874109263657959e-05,
|
| 143 |
+
"loss": 0.2615,
|
| 144 |
+
"step": 900
|
| 145 |
+
},
|
| 146 |
+
{
|
| 147 |
+
"epoch": 1.1282660332541568,
|
| 148 |
+
"grad_norm": 3.425048351287842,
|
| 149 |
+
"learning_rate": 1.247822644497229e-05,
|
| 150 |
+
"loss": 0.247,
|
| 151 |
+
"step": 950
|
| 152 |
+
},
|
| 153 |
+
{
|
| 154 |
+
"epoch": 1.187648456057007,
|
| 155 |
+
"grad_norm": 5.914585113525391,
|
| 156 |
+
"learning_rate": 1.208234362628662e-05,
|
| 157 |
+
"loss": 0.2392,
|
| 158 |
+
"step": 1000
|
| 159 |
+
},
|
| 160 |
+
{
|
| 161 |
+
"epoch": 1.2470308788598574,
|
| 162 |
+
"grad_norm": 5.886987209320068,
|
| 163 |
+
"learning_rate": 1.168646080760095e-05,
|
| 164 |
+
"loss": 0.2574,
|
| 165 |
+
"step": 1050
|
| 166 |
+
},
|
| 167 |
+
{
|
| 168 |
+
"epoch": 1.3064133016627077,
|
| 169 |
+
"grad_norm": 4.573117733001709,
|
| 170 |
+
"learning_rate": 1.1290577988915281e-05,
|
| 171 |
+
"loss": 0.245,
|
| 172 |
+
"step": 1100
|
| 173 |
+
},
|
| 174 |
+
{
|
| 175 |
+
"epoch": 1.365795724465558,
|
| 176 |
+
"grad_norm": 4.632619857788086,
|
| 177 |
+
"learning_rate": 1.0894695170229613e-05,
|
| 178 |
+
"loss": 0.2331,
|
| 179 |
+
"step": 1150
|
| 180 |
+
},
|
| 181 |
+
{
|
| 182 |
+
"epoch": 1.4251781472684084,
|
| 183 |
+
"grad_norm": 6.250412464141846,
|
| 184 |
+
"learning_rate": 1.0498812351543943e-05,
|
| 185 |
+
"loss": 0.2268,
|
| 186 |
+
"step": 1200
|
| 187 |
+
},
|
| 188 |
+
{
|
| 189 |
+
"epoch": 1.484560570071259,
|
| 190 |
+
"grad_norm": 7.933732032775879,
|
| 191 |
+
"learning_rate": 1.0102929532858275e-05,
|
| 192 |
+
"loss": 0.2407,
|
| 193 |
+
"step": 1250
|
| 194 |
+
},
|
| 195 |
+
{
|
| 196 |
+
"epoch": 1.5439429928741093,
|
| 197 |
+
"grad_norm": 5.896273136138916,
|
| 198 |
+
"learning_rate": 9.707046714172605e-06,
|
| 199 |
+
"loss": 0.2447,
|
| 200 |
+
"step": 1300
|
| 201 |
+
},
|
| 202 |
+
{
|
| 203 |
+
"epoch": 1.6033254156769596,
|
| 204 |
+
"grad_norm": 7.057689189910889,
|
| 205 |
+
"learning_rate": 9.311163895486937e-06,
|
| 206 |
+
"loss": 0.2065,
|
| 207 |
+
"step": 1350
|
| 208 |
+
},
|
| 209 |
+
{
|
| 210 |
+
"epoch": 1.66270783847981,
|
| 211 |
+
"grad_norm": 3.920283317565918,
|
| 212 |
+
"learning_rate": 8.915281076801267e-06,
|
| 213 |
+
"loss": 0.2497,
|
| 214 |
+
"step": 1400
|
| 215 |
+
},
|
| 216 |
+
{
|
| 217 |
+
"epoch": 1.7220902612826603,
|
| 218 |
+
"grad_norm": 3.5397119522094727,
|
| 219 |
+
"learning_rate": 8.5193982581156e-06,
|
| 220 |
+
"loss": 0.2097,
|
| 221 |
+
"step": 1450
|
| 222 |
+
},
|
| 223 |
+
{
|
| 224 |
+
"epoch": 1.7814726840855108,
|
| 225 |
+
"grad_norm": 5.607302188873291,
|
| 226 |
+
"learning_rate": 8.12351543942993e-06,
|
| 227 |
+
"loss": 0.2128,
|
| 228 |
+
"step": 1500
|
| 229 |
+
},
|
| 230 |
+
{
|
| 231 |
+
"epoch": 1.8408551068883612,
|
| 232 |
+
"grad_norm": 5.921725749969482,
|
| 233 |
+
"learning_rate": 7.72763262074426e-06,
|
| 234 |
+
"loss": 0.2309,
|
| 235 |
+
"step": 1550
|
| 236 |
+
},
|
| 237 |
+
{
|
| 238 |
+
"epoch": 1.9002375296912115,
|
| 239 |
+
"grad_norm": 3.5897367000579834,
|
| 240 |
+
"learning_rate": 7.331749802058591e-06,
|
| 241 |
+
"loss": 0.225,
|
| 242 |
+
"step": 1600
|
| 243 |
+
},
|
| 244 |
+
{
|
| 245 |
+
"epoch": 1.9596199524940618,
|
| 246 |
+
"grad_norm": 8.864216804504395,
|
| 247 |
+
"learning_rate": 6.935866983372922e-06,
|
| 248 |
+
"loss": 0.2167,
|
| 249 |
+
"step": 1650
|
| 250 |
+
},
|
| 251 |
+
{
|
| 252 |
+
"epoch": 2.0,
|
| 253 |
+
"eval_accuracy": 0.875,
|
| 254 |
+
"eval_loss": 0.36845821142196655,
|
| 255 |
+
"eval_runtime": 2.2567,
|
| 256 |
+
"eval_samples_per_second": 386.402,
|
| 257 |
+
"eval_steps_per_second": 2.659,
|
| 258 |
+
"step": 1684
|
| 259 |
+
},
|
| 260 |
+
{
|
| 261 |
+
"epoch": 2.019002375296912,
|
| 262 |
+
"grad_norm": 5.464975357055664,
|
| 263 |
+
"learning_rate": 6.539984164687253e-06,
|
| 264 |
+
"loss": 0.1996,
|
| 265 |
+
"step": 1700
|
| 266 |
+
},
|
| 267 |
+
{
|
| 268 |
+
"epoch": 2.0783847980997625,
|
| 269 |
+
"grad_norm": 4.667825222015381,
|
| 270 |
+
"learning_rate": 6.1441013460015846e-06,
|
| 271 |
+
"loss": 0.1771,
|
| 272 |
+
"step": 1750
|
| 273 |
+
},
|
| 274 |
+
{
|
| 275 |
+
"epoch": 2.137767220902613,
|
| 276 |
+
"grad_norm": 4.918072700500488,
|
| 277 |
+
"learning_rate": 5.748218527315916e-06,
|
| 278 |
+
"loss": 0.1746,
|
| 279 |
+
"step": 1800
|
| 280 |
+
},
|
| 281 |
+
{
|
| 282 |
+
"epoch": 2.197149643705463,
|
| 283 |
+
"grad_norm": 6.504213809967041,
|
| 284 |
+
"learning_rate": 5.352335708630247e-06,
|
| 285 |
+
"loss": 0.1782,
|
| 286 |
+
"step": 1850
|
| 287 |
+
},
|
| 288 |
+
{
|
| 289 |
+
"epoch": 2.2565320665083135,
|
| 290 |
+
"grad_norm": 7.027709484100342,
|
| 291 |
+
"learning_rate": 4.956452889944577e-06,
|
| 292 |
+
"loss": 0.183,
|
| 293 |
+
"step": 1900
|
| 294 |
+
},
|
| 295 |
+
{
|
| 296 |
+
"epoch": 2.315914489311164,
|
| 297 |
+
"grad_norm": 6.394329071044922,
|
| 298 |
+
"learning_rate": 4.560570071258908e-06,
|
| 299 |
+
"loss": 0.1857,
|
| 300 |
+
"step": 1950
|
| 301 |
+
},
|
| 302 |
+
{
|
| 303 |
+
"epoch": 2.375296912114014,
|
| 304 |
+
"grad_norm": 7.2005510330200195,
|
| 305 |
+
"learning_rate": 4.164687252573238e-06,
|
| 306 |
+
"loss": 0.1888,
|
| 307 |
+
"step": 2000
|
| 308 |
+
},
|
| 309 |
+
{
|
| 310 |
+
"epoch": 2.4346793349168645,
|
| 311 |
+
"grad_norm": 5.697535991668701,
|
| 312 |
+
"learning_rate": 3.7688044338875695e-06,
|
| 313 |
+
"loss": 0.1815,
|
| 314 |
+
"step": 2050
|
| 315 |
+
},
|
| 316 |
+
{
|
| 317 |
+
"epoch": 2.494061757719715,
|
| 318 |
+
"grad_norm": 5.615750789642334,
|
| 319 |
+
"learning_rate": 3.3729216152019006e-06,
|
| 320 |
+
"loss": 0.1566,
|
| 321 |
+
"step": 2100
|
| 322 |
+
},
|
| 323 |
+
{
|
| 324 |
+
"epoch": 2.553444180522565,
|
| 325 |
+
"grad_norm": 4.060975551605225,
|
| 326 |
+
"learning_rate": 2.977038796516231e-06,
|
| 327 |
+
"loss": 0.1818,
|
| 328 |
+
"step": 2150
|
| 329 |
+
},
|
| 330 |
+
{
|
| 331 |
+
"epoch": 2.6128266033254155,
|
| 332 |
+
"grad_norm": 6.223606586456299,
|
| 333 |
+
"learning_rate": 2.581155977830562e-06,
|
| 334 |
+
"loss": 0.1808,
|
| 335 |
+
"step": 2200
|
| 336 |
+
},
|
| 337 |
+
{
|
| 338 |
+
"epoch": 2.6722090261282663,
|
| 339 |
+
"grad_norm": 4.982829570770264,
|
| 340 |
+
"learning_rate": 2.1852731591448932e-06,
|
| 341 |
+
"loss": 0.1733,
|
| 342 |
+
"step": 2250
|
| 343 |
+
},
|
| 344 |
+
{
|
| 345 |
+
"epoch": 2.731591448931116,
|
| 346 |
+
"grad_norm": 7.888515472412109,
|
| 347 |
+
"learning_rate": 1.789390340459224e-06,
|
| 348 |
+
"loss": 0.1671,
|
| 349 |
+
"step": 2300
|
| 350 |
+
},
|
| 351 |
+
{
|
| 352 |
+
"epoch": 2.790973871733967,
|
| 353 |
+
"grad_norm": 5.7161431312561035,
|
| 354 |
+
"learning_rate": 1.3935075217735553e-06,
|
| 355 |
+
"loss": 0.1818,
|
| 356 |
+
"step": 2350
|
| 357 |
+
},
|
| 358 |
+
{
|
| 359 |
+
"epoch": 2.850356294536817,
|
| 360 |
+
"grad_norm": 6.20425271987915,
|
| 361 |
+
"learning_rate": 9.976247030878861e-07,
|
| 362 |
+
"loss": 0.1684,
|
| 363 |
+
"step": 2400
|
| 364 |
+
},
|
| 365 |
+
{
|
| 366 |
+
"epoch": 2.9097387173396676,
|
| 367 |
+
"grad_norm": 10.219502449035645,
|
| 368 |
+
"learning_rate": 6.01741884402217e-07,
|
| 369 |
+
"loss": 0.1656,
|
| 370 |
+
"step": 2450
|
| 371 |
+
},
|
| 372 |
+
{
|
| 373 |
+
"epoch": 2.969121140142518,
|
| 374 |
+
"grad_norm": 5.146057605743408,
|
| 375 |
+
"learning_rate": 2.0585906571654793e-07,
|
| 376 |
+
"loss": 0.1705,
|
| 377 |
+
"step": 2500
|
| 378 |
+
},
|
| 379 |
+
{
|
| 380 |
+
"epoch": 3.0,
|
| 381 |
+
"eval_accuracy": 0.8704128440366973,
|
| 382 |
+
"eval_loss": 0.39564406871795654,
|
| 383 |
+
"eval_runtime": 2.3756,
|
| 384 |
+
"eval_samples_per_second": 367.063,
|
| 385 |
+
"eval_steps_per_second": 2.526,
|
| 386 |
+
"step": 2526
|
| 387 |
+
},
|
| 388 |
+
{
|
| 389 |
+
"epoch": 3.0,
|
| 390 |
+
"step": 2526,
|
| 391 |
+
"total_flos": 6631676508966912.0,
|
| 392 |
+
"train_loss": 0.25985605001260814,
|
| 393 |
+
"train_runtime": 258.2146,
|
| 394 |
+
"train_samples_per_second": 782.477,
|
| 395 |
+
"train_steps_per_second": 9.783
|
| 396 |
+
}
|
| 397 |
+
],
|
| 398 |
+
"logging_steps": 50,
|
| 399 |
+
"max_steps": 2526,
|
| 400 |
+
"num_input_tokens_seen": 0,
|
| 401 |
+
"num_train_epochs": 3,
|
| 402 |
+
"save_steps": 500,
|
| 403 |
+
"stateful_callbacks": {
|
| 404 |
+
"TrainerControl": {
|
| 405 |
+
"args": {
|
| 406 |
+
"should_epoch_stop": false,
|
| 407 |
+
"should_evaluate": false,
|
| 408 |
+
"should_log": false,
|
| 409 |
+
"should_save": false,
|
| 410 |
+
"should_training_stop": false
|
| 411 |
+
},
|
| 412 |
+
"attributes": {}
|
| 413 |
+
}
|
| 414 |
+
},
|
| 415 |
+
"total_flos": 6631676508966912.0,
|
| 416 |
+
"train_batch_size": 80,
|
| 417 |
+
"trial_name": null,
|
| 418 |
+
"trial_params": null
|
| 419 |
+
}
|
mbert_paper_metrics/result_ablation_mbert_paper/sst2/mBERT/max/seed_43/all_results.json
ADDED
|
@@ -0,0 +1,15 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 3.0,
|
| 3 |
+
"eval_accuracy": 0.8807339449541285,
|
| 4 |
+
"eval_loss": 0.3865836262702942,
|
| 5 |
+
"eval_runtime": 2.4044,
|
| 6 |
+
"eval_samples": 872,
|
| 7 |
+
"eval_samples_per_second": 362.674,
|
| 8 |
+
"eval_steps_per_second": 2.495,
|
| 9 |
+
"total_flos": 6637229322706944.0,
|
| 10 |
+
"train_loss": 0.2557056057764629,
|
| 11 |
+
"train_runtime": 269.9813,
|
| 12 |
+
"train_samples": 67349,
|
| 13 |
+
"train_samples_per_second": 748.374,
|
| 14 |
+
"train_steps_per_second": 9.356
|
| 15 |
+
}
|
mbert_paper_metrics/result_ablation_mbert_paper/sst2/mBERT/max/seed_43/eval_results.json
ADDED
|
@@ -0,0 +1,9 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 3.0,
|
| 3 |
+
"eval_accuracy": 0.8807339449541285,
|
| 4 |
+
"eval_loss": 0.3865836262702942,
|
| 5 |
+
"eval_runtime": 2.4044,
|
| 6 |
+
"eval_samples": 872,
|
| 7 |
+
"eval_samples_per_second": 362.674,
|
| 8 |
+
"eval_steps_per_second": 2.495
|
| 9 |
+
}
|
mbert_paper_metrics/result_ablation_mbert_paper/sst2/mBERT/max/seed_43/run_command.txt
ADDED
|
@@ -0,0 +1 @@
|
|
|
|
|
|
|
| 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 sst2 --max_seq_length 256 --do_train --do_eval --evaluation_strategy epoch --save_strategy no --logging_strategy steps --logging_steps 50 --report_to none --num_train_epochs 3 --learning_rate 2e-05 --per_device_train_batch_size 80 --per_device_eval_batch_size 168 --gradient_accumulation_steps 1 --dataloader_num_workers 2 --seed 43 --output_dir result_ablation_mbert_paper/sst2/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 8 --head_dim 32 --classifier_hidden_dim 512 --head_dropout 0.05
|
mbert_paper_metrics/result_ablation_mbert_paper/sst2/mBERT/max/seed_43/train_results.json
ADDED
|
@@ -0,0 +1,9 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 3.0,
|
| 3 |
+
"total_flos": 6637229322706944.0,
|
| 4 |
+
"train_loss": 0.2557056057764629,
|
| 5 |
+
"train_runtime": 269.9813,
|
| 6 |
+
"train_samples": 67349,
|
| 7 |
+
"train_samples_per_second": 748.374,
|
| 8 |
+
"train_steps_per_second": 9.356
|
| 9 |
+
}
|
mbert_paper_metrics/result_ablation_mbert_paper/sst2/mBERT/max/seed_43/trainer_state.json
ADDED
|
@@ -0,0 +1,419 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_metric": null,
|
| 3 |
+
"best_model_checkpoint": null,
|
| 4 |
+
"epoch": 3.0,
|
| 5 |
+
"eval_steps": 500,
|
| 6 |
+
"global_step": 2526,
|
| 7 |
+
"is_hyper_param_search": false,
|
| 8 |
+
"is_local_process_zero": true,
|
| 9 |
+
"is_world_process_zero": true,
|
| 10 |
+
"log_history": [
|
| 11 |
+
{
|
| 12 |
+
"epoch": 0.05938242280285035,
|
| 13 |
+
"grad_norm": 4.180697441101074,
|
| 14 |
+
"learning_rate": 1.9604117181314332e-05,
|
| 15 |
+
"loss": 0.5702,
|
| 16 |
+
"step": 50
|
| 17 |
+
},
|
| 18 |
+
{
|
| 19 |
+
"epoch": 0.1187648456057007,
|
| 20 |
+
"grad_norm": 4.17879581451416,
|
| 21 |
+
"learning_rate": 1.9208234362628665e-05,
|
| 22 |
+
"loss": 0.4617,
|
| 23 |
+
"step": 100
|
| 24 |
+
},
|
| 25 |
+
{
|
| 26 |
+
"epoch": 0.17814726840855108,
|
| 27 |
+
"grad_norm": 6.864574909210205,
|
| 28 |
+
"learning_rate": 1.8812351543942996e-05,
|
| 29 |
+
"loss": 0.4285,
|
| 30 |
+
"step": 150
|
| 31 |
+
},
|
| 32 |
+
{
|
| 33 |
+
"epoch": 0.2375296912114014,
|
| 34 |
+
"grad_norm": 4.7572736740112305,
|
| 35 |
+
"learning_rate": 1.8416468725257326e-05,
|
| 36 |
+
"loss": 0.4196,
|
| 37 |
+
"step": 200
|
| 38 |
+
},
|
| 39 |
+
{
|
| 40 |
+
"epoch": 0.29691211401425177,
|
| 41 |
+
"grad_norm": 3.6587648391723633,
|
| 42 |
+
"learning_rate": 1.8020585906571656e-05,
|
| 43 |
+
"loss": 0.4115,
|
| 44 |
+
"step": 250
|
| 45 |
+
},
|
| 46 |
+
{
|
| 47 |
+
"epoch": 0.35629453681710216,
|
| 48 |
+
"grad_norm": 4.370148658752441,
|
| 49 |
+
"learning_rate": 1.762470308788599e-05,
|
| 50 |
+
"loss": 0.372,
|
| 51 |
+
"step": 300
|
| 52 |
+
},
|
| 53 |
+
{
|
| 54 |
+
"epoch": 0.4156769596199525,
|
| 55 |
+
"grad_norm": 4.285806655883789,
|
| 56 |
+
"learning_rate": 1.7228820269200316e-05,
|
| 57 |
+
"loss": 0.353,
|
| 58 |
+
"step": 350
|
| 59 |
+
},
|
| 60 |
+
{
|
| 61 |
+
"epoch": 0.4750593824228028,
|
| 62 |
+
"grad_norm": 4.065441608428955,
|
| 63 |
+
"learning_rate": 1.6832937450514647e-05,
|
| 64 |
+
"loss": 0.3549,
|
| 65 |
+
"step": 400
|
| 66 |
+
},
|
| 67 |
+
{
|
| 68 |
+
"epoch": 0.5344418052256532,
|
| 69 |
+
"grad_norm": 5.238286972045898,
|
| 70 |
+
"learning_rate": 1.643705463182898e-05,
|
| 71 |
+
"loss": 0.3419,
|
| 72 |
+
"step": 450
|
| 73 |
+
},
|
| 74 |
+
{
|
| 75 |
+
"epoch": 0.5938242280285035,
|
| 76 |
+
"grad_norm": 3.9033684730529785,
|
| 77 |
+
"learning_rate": 1.604117181314331e-05,
|
| 78 |
+
"loss": 0.33,
|
| 79 |
+
"step": 500
|
| 80 |
+
},
|
| 81 |
+
{
|
| 82 |
+
"epoch": 0.6532066508313539,
|
| 83 |
+
"grad_norm": 6.779426097869873,
|
| 84 |
+
"learning_rate": 1.564528899445764e-05,
|
| 85 |
+
"loss": 0.3176,
|
| 86 |
+
"step": 550
|
| 87 |
+
},
|
| 88 |
+
{
|
| 89 |
+
"epoch": 0.7125890736342043,
|
| 90 |
+
"grad_norm": 4.882399559020996,
|
| 91 |
+
"learning_rate": 1.5249406175771972e-05,
|
| 92 |
+
"loss": 0.3088,
|
| 93 |
+
"step": 600
|
| 94 |
+
},
|
| 95 |
+
{
|
| 96 |
+
"epoch": 0.7719714964370546,
|
| 97 |
+
"grad_norm": 5.467887878417969,
|
| 98 |
+
"learning_rate": 1.4853523357086304e-05,
|
| 99 |
+
"loss": 0.3065,
|
| 100 |
+
"step": 650
|
| 101 |
+
},
|
| 102 |
+
{
|
| 103 |
+
"epoch": 0.831353919239905,
|
| 104 |
+
"grad_norm": 4.645595073699951,
|
| 105 |
+
"learning_rate": 1.4457640538400634e-05,
|
| 106 |
+
"loss": 0.3104,
|
| 107 |
+
"step": 700
|
| 108 |
+
},
|
| 109 |
+
{
|
| 110 |
+
"epoch": 0.8907363420427553,
|
| 111 |
+
"grad_norm": 4.219912528991699,
|
| 112 |
+
"learning_rate": 1.4061757719714966e-05,
|
| 113 |
+
"loss": 0.3117,
|
| 114 |
+
"step": 750
|
| 115 |
+
},
|
| 116 |
+
{
|
| 117 |
+
"epoch": 0.9501187648456056,
|
| 118 |
+
"grad_norm": 5.764337062835693,
|
| 119 |
+
"learning_rate": 1.3665874901029297e-05,
|
| 120 |
+
"loss": 0.2994,
|
| 121 |
+
"step": 800
|
| 122 |
+
},
|
| 123 |
+
{
|
| 124 |
+
"epoch": 1.0,
|
| 125 |
+
"eval_accuracy": 0.875,
|
| 126 |
+
"eval_loss": 0.33486175537109375,
|
| 127 |
+
"eval_runtime": 2.3074,
|
| 128 |
+
"eval_samples_per_second": 377.906,
|
| 129 |
+
"eval_steps_per_second": 2.6,
|
| 130 |
+
"step": 842
|
| 131 |
+
},
|
| 132 |
+
{
|
| 133 |
+
"epoch": 1.009501187648456,
|
| 134 |
+
"grad_norm": 4.989134788513184,
|
| 135 |
+
"learning_rate": 1.3269992082343628e-05,
|
| 136 |
+
"loss": 0.2799,
|
| 137 |
+
"step": 850
|
| 138 |
+
},
|
| 139 |
+
{
|
| 140 |
+
"epoch": 1.0688836104513064,
|
| 141 |
+
"grad_norm": 5.427011489868164,
|
| 142 |
+
"learning_rate": 1.2874109263657959e-05,
|
| 143 |
+
"loss": 0.2446,
|
| 144 |
+
"step": 900
|
| 145 |
+
},
|
| 146 |
+
{
|
| 147 |
+
"epoch": 1.1282660332541568,
|
| 148 |
+
"grad_norm": 4.836376190185547,
|
| 149 |
+
"learning_rate": 1.247822644497229e-05,
|
| 150 |
+
"loss": 0.2341,
|
| 151 |
+
"step": 950
|
| 152 |
+
},
|
| 153 |
+
{
|
| 154 |
+
"epoch": 1.187648456057007,
|
| 155 |
+
"grad_norm": 4.55562686920166,
|
| 156 |
+
"learning_rate": 1.208234362628662e-05,
|
| 157 |
+
"loss": 0.2476,
|
| 158 |
+
"step": 1000
|
| 159 |
+
},
|
| 160 |
+
{
|
| 161 |
+
"epoch": 1.2470308788598574,
|
| 162 |
+
"grad_norm": 6.445218563079834,
|
| 163 |
+
"learning_rate": 1.168646080760095e-05,
|
| 164 |
+
"loss": 0.241,
|
| 165 |
+
"step": 1050
|
| 166 |
+
},
|
| 167 |
+
{
|
| 168 |
+
"epoch": 1.3064133016627077,
|
| 169 |
+
"grad_norm": 6.573553562164307,
|
| 170 |
+
"learning_rate": 1.1290577988915281e-05,
|
| 171 |
+
"loss": 0.2425,
|
| 172 |
+
"step": 1100
|
| 173 |
+
},
|
| 174 |
+
{
|
| 175 |
+
"epoch": 1.365795724465558,
|
| 176 |
+
"grad_norm": 7.740910053253174,
|
| 177 |
+
"learning_rate": 1.0894695170229613e-05,
|
| 178 |
+
"loss": 0.2399,
|
| 179 |
+
"step": 1150
|
| 180 |
+
},
|
| 181 |
+
{
|
| 182 |
+
"epoch": 1.4251781472684084,
|
| 183 |
+
"grad_norm": 5.468100547790527,
|
| 184 |
+
"learning_rate": 1.0498812351543943e-05,
|
| 185 |
+
"loss": 0.2445,
|
| 186 |
+
"step": 1200
|
| 187 |
+
},
|
| 188 |
+
{
|
| 189 |
+
"epoch": 1.484560570071259,
|
| 190 |
+
"grad_norm": 7.666905879974365,
|
| 191 |
+
"learning_rate": 1.0102929532858275e-05,
|
| 192 |
+
"loss": 0.224,
|
| 193 |
+
"step": 1250
|
| 194 |
+
},
|
| 195 |
+
{
|
| 196 |
+
"epoch": 1.5439429928741093,
|
| 197 |
+
"grad_norm": 3.8822779655456543,
|
| 198 |
+
"learning_rate": 9.707046714172605e-06,
|
| 199 |
+
"loss": 0.2445,
|
| 200 |
+
"step": 1300
|
| 201 |
+
},
|
| 202 |
+
{
|
| 203 |
+
"epoch": 1.6033254156769596,
|
| 204 |
+
"grad_norm": 5.127964019775391,
|
| 205 |
+
"learning_rate": 9.311163895486937e-06,
|
| 206 |
+
"loss": 0.2326,
|
| 207 |
+
"step": 1350
|
| 208 |
+
},
|
| 209 |
+
{
|
| 210 |
+
"epoch": 1.66270783847981,
|
| 211 |
+
"grad_norm": 6.518642902374268,
|
| 212 |
+
"learning_rate": 8.915281076801267e-06,
|
| 213 |
+
"loss": 0.2478,
|
| 214 |
+
"step": 1400
|
| 215 |
+
},
|
| 216 |
+
{
|
| 217 |
+
"epoch": 1.7220902612826603,
|
| 218 |
+
"grad_norm": 5.288575172424316,
|
| 219 |
+
"learning_rate": 8.5193982581156e-06,
|
| 220 |
+
"loss": 0.219,
|
| 221 |
+
"step": 1450
|
| 222 |
+
},
|
| 223 |
+
{
|
| 224 |
+
"epoch": 1.7814726840855108,
|
| 225 |
+
"grad_norm": 5.339145183563232,
|
| 226 |
+
"learning_rate": 8.12351543942993e-06,
|
| 227 |
+
"loss": 0.2299,
|
| 228 |
+
"step": 1500
|
| 229 |
+
},
|
| 230 |
+
{
|
| 231 |
+
"epoch": 1.8408551068883612,
|
| 232 |
+
"grad_norm": 7.03946590423584,
|
| 233 |
+
"learning_rate": 7.72763262074426e-06,
|
| 234 |
+
"loss": 0.2117,
|
| 235 |
+
"step": 1550
|
| 236 |
+
},
|
| 237 |
+
{
|
| 238 |
+
"epoch": 1.9002375296912115,
|
| 239 |
+
"grad_norm": 5.588253974914551,
|
| 240 |
+
"learning_rate": 7.331749802058591e-06,
|
| 241 |
+
"loss": 0.2107,
|
| 242 |
+
"step": 1600
|
| 243 |
+
},
|
| 244 |
+
{
|
| 245 |
+
"epoch": 1.9596199524940618,
|
| 246 |
+
"grad_norm": 5.119135856628418,
|
| 247 |
+
"learning_rate": 6.935866983372922e-06,
|
| 248 |
+
"loss": 0.2128,
|
| 249 |
+
"step": 1650
|
| 250 |
+
},
|
| 251 |
+
{
|
| 252 |
+
"epoch": 2.0,
|
| 253 |
+
"eval_accuracy": 0.8807339449541285,
|
| 254 |
+
"eval_loss": 0.359303742647171,
|
| 255 |
+
"eval_runtime": 2.1792,
|
| 256 |
+
"eval_samples_per_second": 400.139,
|
| 257 |
+
"eval_steps_per_second": 2.753,
|
| 258 |
+
"step": 1684
|
| 259 |
+
},
|
| 260 |
+
{
|
| 261 |
+
"epoch": 2.019002375296912,
|
| 262 |
+
"grad_norm": 5.391238689422607,
|
| 263 |
+
"learning_rate": 6.539984164687253e-06,
|
| 264 |
+
"loss": 0.1968,
|
| 265 |
+
"step": 1700
|
| 266 |
+
},
|
| 267 |
+
{
|
| 268 |
+
"epoch": 2.0783847980997625,
|
| 269 |
+
"grad_norm": 6.429093837738037,
|
| 270 |
+
"learning_rate": 6.1441013460015846e-06,
|
| 271 |
+
"loss": 0.1659,
|
| 272 |
+
"step": 1750
|
| 273 |
+
},
|
| 274 |
+
{
|
| 275 |
+
"epoch": 2.137767220902613,
|
| 276 |
+
"grad_norm": 5.3665008544921875,
|
| 277 |
+
"learning_rate": 5.748218527315916e-06,
|
| 278 |
+
"loss": 0.1735,
|
| 279 |
+
"step": 1800
|
| 280 |
+
},
|
| 281 |
+
{
|
| 282 |
+
"epoch": 2.197149643705463,
|
| 283 |
+
"grad_norm": 7.636340618133545,
|
| 284 |
+
"learning_rate": 5.352335708630247e-06,
|
| 285 |
+
"loss": 0.1791,
|
| 286 |
+
"step": 1850
|
| 287 |
+
},
|
| 288 |
+
{
|
| 289 |
+
"epoch": 2.2565320665083135,
|
| 290 |
+
"grad_norm": 7.158308029174805,
|
| 291 |
+
"learning_rate": 4.956452889944577e-06,
|
| 292 |
+
"loss": 0.1669,
|
| 293 |
+
"step": 1900
|
| 294 |
+
},
|
| 295 |
+
{
|
| 296 |
+
"epoch": 2.315914489311164,
|
| 297 |
+
"grad_norm": 7.830133438110352,
|
| 298 |
+
"learning_rate": 4.560570071258908e-06,
|
| 299 |
+
"loss": 0.1753,
|
| 300 |
+
"step": 1950
|
| 301 |
+
},
|
| 302 |
+
{
|
| 303 |
+
"epoch": 2.375296912114014,
|
| 304 |
+
"grad_norm": 5.292535781860352,
|
| 305 |
+
"learning_rate": 4.164687252573238e-06,
|
| 306 |
+
"loss": 0.1636,
|
| 307 |
+
"step": 2000
|
| 308 |
+
},
|
| 309 |
+
{
|
| 310 |
+
"epoch": 2.4346793349168645,
|
| 311 |
+
"grad_norm": 5.023130416870117,
|
| 312 |
+
"learning_rate": 3.7688044338875695e-06,
|
| 313 |
+
"loss": 0.1671,
|
| 314 |
+
"step": 2050
|
| 315 |
+
},
|
| 316 |
+
{
|
| 317 |
+
"epoch": 2.494061757719715,
|
| 318 |
+
"grad_norm": 5.847843170166016,
|
| 319 |
+
"learning_rate": 3.3729216152019006e-06,
|
| 320 |
+
"loss": 0.1714,
|
| 321 |
+
"step": 2100
|
| 322 |
+
},
|
| 323 |
+
{
|
| 324 |
+
"epoch": 2.553444180522565,
|
| 325 |
+
"grad_norm": 4.666432857513428,
|
| 326 |
+
"learning_rate": 2.977038796516231e-06,
|
| 327 |
+
"loss": 0.1798,
|
| 328 |
+
"step": 2150
|
| 329 |
+
},
|
| 330 |
+
{
|
| 331 |
+
"epoch": 2.6128266033254155,
|
| 332 |
+
"grad_norm": 4.751523971557617,
|
| 333 |
+
"learning_rate": 2.581155977830562e-06,
|
| 334 |
+
"loss": 0.1719,
|
| 335 |
+
"step": 2200
|
| 336 |
+
},
|
| 337 |
+
{
|
| 338 |
+
"epoch": 2.6722090261282663,
|
| 339 |
+
"grad_norm": 8.068571090698242,
|
| 340 |
+
"learning_rate": 2.1852731591448932e-06,
|
| 341 |
+
"loss": 0.154,
|
| 342 |
+
"step": 2250
|
| 343 |
+
},
|
| 344 |
+
{
|
| 345 |
+
"epoch": 2.731591448931116,
|
| 346 |
+
"grad_norm": 6.897275924682617,
|
| 347 |
+
"learning_rate": 1.789390340459224e-06,
|
| 348 |
+
"loss": 0.1824,
|
| 349 |
+
"step": 2300
|
| 350 |
+
},
|
| 351 |
+
{
|
| 352 |
+
"epoch": 2.790973871733967,
|
| 353 |
+
"grad_norm": 7.1154303550720215,
|
| 354 |
+
"learning_rate": 1.3935075217735553e-06,
|
| 355 |
+
"loss": 0.1705,
|
| 356 |
+
"step": 2350
|
| 357 |
+
},
|
| 358 |
+
{
|
| 359 |
+
"epoch": 2.850356294536817,
|
| 360 |
+
"grad_norm": 6.588690280914307,
|
| 361 |
+
"learning_rate": 9.976247030878861e-07,
|
| 362 |
+
"loss": 0.1684,
|
| 363 |
+
"step": 2400
|
| 364 |
+
},
|
| 365 |
+
{
|
| 366 |
+
"epoch": 2.9097387173396676,
|
| 367 |
+
"grad_norm": 5.93678617477417,
|
| 368 |
+
"learning_rate": 6.01741884402217e-07,
|
| 369 |
+
"loss": 0.1678,
|
| 370 |
+
"step": 2450
|
| 371 |
+
},
|
| 372 |
+
{
|
| 373 |
+
"epoch": 2.969121140142518,
|
| 374 |
+
"grad_norm": 3.7040066719055176,
|
| 375 |
+
"learning_rate": 2.0585906571654793e-07,
|
| 376 |
+
"loss": 0.1679,
|
| 377 |
+
"step": 2500
|
| 378 |
+
},
|
| 379 |
+
{
|
| 380 |
+
"epoch": 3.0,
|
| 381 |
+
"eval_accuracy": 0.8807339449541285,
|
| 382 |
+
"eval_loss": 0.3865836262702942,
|
| 383 |
+
"eval_runtime": 2.4133,
|
| 384 |
+
"eval_samples_per_second": 361.327,
|
| 385 |
+
"eval_steps_per_second": 2.486,
|
| 386 |
+
"step": 2526
|
| 387 |
+
},
|
| 388 |
+
{
|
| 389 |
+
"epoch": 3.0,
|
| 390 |
+
"step": 2526,
|
| 391 |
+
"total_flos": 6637229322706944.0,
|
| 392 |
+
"train_loss": 0.2557056057764629,
|
| 393 |
+
"train_runtime": 269.9813,
|
| 394 |
+
"train_samples_per_second": 748.374,
|
| 395 |
+
"train_steps_per_second": 9.356
|
| 396 |
+
}
|
| 397 |
+
],
|
| 398 |
+
"logging_steps": 50,
|
| 399 |
+
"max_steps": 2526,
|
| 400 |
+
"num_input_tokens_seen": 0,
|
| 401 |
+
"num_train_epochs": 3,
|
| 402 |
+
"save_steps": 500,
|
| 403 |
+
"stateful_callbacks": {
|
| 404 |
+
"TrainerControl": {
|
| 405 |
+
"args": {
|
| 406 |
+
"should_epoch_stop": false,
|
| 407 |
+
"should_evaluate": false,
|
| 408 |
+
"should_log": false,
|
| 409 |
+
"should_save": false,
|
| 410 |
+
"should_training_stop": false
|
| 411 |
+
},
|
| 412 |
+
"attributes": {}
|
| 413 |
+
}
|
| 414 |
+
},
|
| 415 |
+
"total_flos": 6637229322706944.0,
|
| 416 |
+
"train_batch_size": 80,
|
| 417 |
+
"trial_name": null,
|
| 418 |
+
"trial_params": null
|
| 419 |
+
}
|
mbert_paper_metrics/result_ablation_mbert_paper/sst2/mBERT/max/seed_44/all_results.json
ADDED
|
@@ -0,0 +1,15 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 3.0,
|
| 3 |
+
"eval_accuracy": 0.8727064220183486,
|
| 4 |
+
"eval_loss": 0.3852188289165497,
|
| 5 |
+
"eval_runtime": 1.9753,
|
| 6 |
+
"eval_samples": 872,
|
| 7 |
+
"eval_samples_per_second": 441.45,
|
| 8 |
+
"eval_steps_per_second": 3.038,
|
| 9 |
+
"total_flos": 6633975382115328.0,
|
| 10 |
+
"train_loss": 0.25833009861804906,
|
| 11 |
+
"train_runtime": 268.2911,
|
| 12 |
+
"train_samples": 67349,
|
| 13 |
+
"train_samples_per_second": 753.089,
|
| 14 |
+
"train_steps_per_second": 9.415
|
| 15 |
+
}
|
mbert_paper_metrics/result_ablation_mbert_paper/sst2/mBERT/max/seed_44/eval_results.json
ADDED
|
@@ -0,0 +1,9 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 3.0,
|
| 3 |
+
"eval_accuracy": 0.8727064220183486,
|
| 4 |
+
"eval_loss": 0.3852188289165497,
|
| 5 |
+
"eval_runtime": 1.9753,
|
| 6 |
+
"eval_samples": 872,
|
| 7 |
+
"eval_samples_per_second": 441.45,
|
| 8 |
+
"eval_steps_per_second": 3.038
|
| 9 |
+
}
|
mbert_paper_metrics/result_ablation_mbert_paper/sst2/mBERT/max/seed_44/run_command.txt
ADDED
|
@@ -0,0 +1 @@
|
|
|
|
|
|
|
| 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 sst2 --max_seq_length 256 --do_train --do_eval --evaluation_strategy epoch --save_strategy no --logging_strategy steps --logging_steps 50 --report_to none --num_train_epochs 3 --learning_rate 2e-05 --per_device_train_batch_size 80 --per_device_eval_batch_size 168 --gradient_accumulation_steps 1 --dataloader_num_workers 2 --seed 44 --output_dir result_ablation_mbert_paper/sst2/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 8 --head_dim 32 --classifier_hidden_dim 512 --head_dropout 0.05
|
mbert_paper_metrics/result_ablation_mbert_paper/sst2/mBERT/max/seed_44/train_results.json
ADDED
|
@@ -0,0 +1,9 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 3.0,
|
| 3 |
+
"total_flos": 6633975382115328.0,
|
| 4 |
+
"train_loss": 0.25833009861804906,
|
| 5 |
+
"train_runtime": 268.2911,
|
| 6 |
+
"train_samples": 67349,
|
| 7 |
+
"train_samples_per_second": 753.089,
|
| 8 |
+
"train_steps_per_second": 9.415
|
| 9 |
+
}
|
mbert_paper_metrics/result_ablation_mbert_paper/sst2/mBERT/max/seed_44/trainer_state.json
ADDED
|
@@ -0,0 +1,419 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_metric": null,
|
| 3 |
+
"best_model_checkpoint": null,
|
| 4 |
+
"epoch": 3.0,
|
| 5 |
+
"eval_steps": 500,
|
| 6 |
+
"global_step": 2526,
|
| 7 |
+
"is_hyper_param_search": false,
|
| 8 |
+
"is_local_process_zero": true,
|
| 9 |
+
"is_world_process_zero": true,
|
| 10 |
+
"log_history": [
|
| 11 |
+
{
|
| 12 |
+
"epoch": 0.05938242280285035,
|
| 13 |
+
"grad_norm": 3.6384241580963135,
|
| 14 |
+
"learning_rate": 1.9604117181314332e-05,
|
| 15 |
+
"loss": 0.595,
|
| 16 |
+
"step": 50
|
| 17 |
+
},
|
| 18 |
+
{
|
| 19 |
+
"epoch": 0.1187648456057007,
|
| 20 |
+
"grad_norm": 3.0850143432617188,
|
| 21 |
+
"learning_rate": 1.9208234362628665e-05,
|
| 22 |
+
"loss": 0.4815,
|
| 23 |
+
"step": 100
|
| 24 |
+
},
|
| 25 |
+
{
|
| 26 |
+
"epoch": 0.17814726840855108,
|
| 27 |
+
"grad_norm": 4.585625171661377,
|
| 28 |
+
"learning_rate": 1.8812351543942996e-05,
|
| 29 |
+
"loss": 0.4325,
|
| 30 |
+
"step": 150
|
| 31 |
+
},
|
| 32 |
+
{
|
| 33 |
+
"epoch": 0.2375296912114014,
|
| 34 |
+
"grad_norm": 3.6793177127838135,
|
| 35 |
+
"learning_rate": 1.8416468725257326e-05,
|
| 36 |
+
"loss": 0.4042,
|
| 37 |
+
"step": 200
|
| 38 |
+
},
|
| 39 |
+
{
|
| 40 |
+
"epoch": 0.29691211401425177,
|
| 41 |
+
"grad_norm": 4.163187026977539,
|
| 42 |
+
"learning_rate": 1.8020585906571656e-05,
|
| 43 |
+
"loss": 0.392,
|
| 44 |
+
"step": 250
|
| 45 |
+
},
|
| 46 |
+
{
|
| 47 |
+
"epoch": 0.35629453681710216,
|
| 48 |
+
"grad_norm": 5.678406715393066,
|
| 49 |
+
"learning_rate": 1.762470308788599e-05,
|
| 50 |
+
"loss": 0.3755,
|
| 51 |
+
"step": 300
|
| 52 |
+
},
|
| 53 |
+
{
|
| 54 |
+
"epoch": 0.4156769596199525,
|
| 55 |
+
"grad_norm": 5.400622844696045,
|
| 56 |
+
"learning_rate": 1.7228820269200316e-05,
|
| 57 |
+
"loss": 0.3701,
|
| 58 |
+
"step": 350
|
| 59 |
+
},
|
| 60 |
+
{
|
| 61 |
+
"epoch": 0.4750593824228028,
|
| 62 |
+
"grad_norm": 4.563175678253174,
|
| 63 |
+
"learning_rate": 1.6832937450514647e-05,
|
| 64 |
+
"loss": 0.3507,
|
| 65 |
+
"step": 400
|
| 66 |
+
},
|
| 67 |
+
{
|
| 68 |
+
"epoch": 0.5344418052256532,
|
| 69 |
+
"grad_norm": 6.190798759460449,
|
| 70 |
+
"learning_rate": 1.643705463182898e-05,
|
| 71 |
+
"loss": 0.334,
|
| 72 |
+
"step": 450
|
| 73 |
+
},
|
| 74 |
+
{
|
| 75 |
+
"epoch": 0.5938242280285035,
|
| 76 |
+
"grad_norm": 8.314579010009766,
|
| 77 |
+
"learning_rate": 1.604117181314331e-05,
|
| 78 |
+
"loss": 0.3424,
|
| 79 |
+
"step": 500
|
| 80 |
+
},
|
| 81 |
+
{
|
| 82 |
+
"epoch": 0.6532066508313539,
|
| 83 |
+
"grad_norm": 5.053552150726318,
|
| 84 |
+
"learning_rate": 1.564528899445764e-05,
|
| 85 |
+
"loss": 0.3315,
|
| 86 |
+
"step": 550
|
| 87 |
+
},
|
| 88 |
+
{
|
| 89 |
+
"epoch": 0.7125890736342043,
|
| 90 |
+
"grad_norm": 6.1065568923950195,
|
| 91 |
+
"learning_rate": 1.5249406175771972e-05,
|
| 92 |
+
"loss": 0.3206,
|
| 93 |
+
"step": 600
|
| 94 |
+
},
|
| 95 |
+
{
|
| 96 |
+
"epoch": 0.7719714964370546,
|
| 97 |
+
"grad_norm": 4.3557305335998535,
|
| 98 |
+
"learning_rate": 1.4853523357086304e-05,
|
| 99 |
+
"loss": 0.3165,
|
| 100 |
+
"step": 650
|
| 101 |
+
},
|
| 102 |
+
{
|
| 103 |
+
"epoch": 0.831353919239905,
|
| 104 |
+
"grad_norm": 3.2330002784729004,
|
| 105 |
+
"learning_rate": 1.4457640538400634e-05,
|
| 106 |
+
"loss": 0.3059,
|
| 107 |
+
"step": 700
|
| 108 |
+
},
|
| 109 |
+
{
|
| 110 |
+
"epoch": 0.8907363420427553,
|
| 111 |
+
"grad_norm": 7.250136375427246,
|
| 112 |
+
"learning_rate": 1.4061757719714966e-05,
|
| 113 |
+
"loss": 0.3011,
|
| 114 |
+
"step": 750
|
| 115 |
+
},
|
| 116 |
+
{
|
| 117 |
+
"epoch": 0.9501187648456056,
|
| 118 |
+
"grad_norm": 4.757879257202148,
|
| 119 |
+
"learning_rate": 1.3665874901029297e-05,
|
| 120 |
+
"loss": 0.3083,
|
| 121 |
+
"step": 800
|
| 122 |
+
},
|
| 123 |
+
{
|
| 124 |
+
"epoch": 1.0,
|
| 125 |
+
"eval_accuracy": 0.8681192660550459,
|
| 126 |
+
"eval_loss": 0.35066020488739014,
|
| 127 |
+
"eval_runtime": 2.4524,
|
| 128 |
+
"eval_samples_per_second": 355.572,
|
| 129 |
+
"eval_steps_per_second": 2.447,
|
| 130 |
+
"step": 842
|
| 131 |
+
},
|
| 132 |
+
{
|
| 133 |
+
"epoch": 1.009501187648456,
|
| 134 |
+
"grad_norm": 9.448796272277832,
|
| 135 |
+
"learning_rate": 1.3269992082343628e-05,
|
| 136 |
+
"loss": 0.2761,
|
| 137 |
+
"step": 850
|
| 138 |
+
},
|
| 139 |
+
{
|
| 140 |
+
"epoch": 1.0688836104513064,
|
| 141 |
+
"grad_norm": 6.732472896575928,
|
| 142 |
+
"learning_rate": 1.2874109263657959e-05,
|
| 143 |
+
"loss": 0.2443,
|
| 144 |
+
"step": 900
|
| 145 |
+
},
|
| 146 |
+
{
|
| 147 |
+
"epoch": 1.1282660332541568,
|
| 148 |
+
"grad_norm": 4.304388999938965,
|
| 149 |
+
"learning_rate": 1.247822644497229e-05,
|
| 150 |
+
"loss": 0.2332,
|
| 151 |
+
"step": 950
|
| 152 |
+
},
|
| 153 |
+
{
|
| 154 |
+
"epoch": 1.187648456057007,
|
| 155 |
+
"grad_norm": 3.327448606491089,
|
| 156 |
+
"learning_rate": 1.208234362628662e-05,
|
| 157 |
+
"loss": 0.2442,
|
| 158 |
+
"step": 1000
|
| 159 |
+
},
|
| 160 |
+
{
|
| 161 |
+
"epoch": 1.2470308788598574,
|
| 162 |
+
"grad_norm": 4.184940338134766,
|
| 163 |
+
"learning_rate": 1.168646080760095e-05,
|
| 164 |
+
"loss": 0.2372,
|
| 165 |
+
"step": 1050
|
| 166 |
+
},
|
| 167 |
+
{
|
| 168 |
+
"epoch": 1.3064133016627077,
|
| 169 |
+
"grad_norm": 4.874513149261475,
|
| 170 |
+
"learning_rate": 1.1290577988915281e-05,
|
| 171 |
+
"loss": 0.2441,
|
| 172 |
+
"step": 1100
|
| 173 |
+
},
|
| 174 |
+
{
|
| 175 |
+
"epoch": 1.365795724465558,
|
| 176 |
+
"grad_norm": 5.032442569732666,
|
| 177 |
+
"learning_rate": 1.0894695170229613e-05,
|
| 178 |
+
"loss": 0.2498,
|
| 179 |
+
"step": 1150
|
| 180 |
+
},
|
| 181 |
+
{
|
| 182 |
+
"epoch": 1.4251781472684084,
|
| 183 |
+
"grad_norm": 4.776379108428955,
|
| 184 |
+
"learning_rate": 1.0498812351543943e-05,
|
| 185 |
+
"loss": 0.2425,
|
| 186 |
+
"step": 1200
|
| 187 |
+
},
|
| 188 |
+
{
|
| 189 |
+
"epoch": 1.484560570071259,
|
| 190 |
+
"grad_norm": 4.78900671005249,
|
| 191 |
+
"learning_rate": 1.0102929532858275e-05,
|
| 192 |
+
"loss": 0.2292,
|
| 193 |
+
"step": 1250
|
| 194 |
+
},
|
| 195 |
+
{
|
| 196 |
+
"epoch": 1.5439429928741093,
|
| 197 |
+
"grad_norm": 7.3016037940979,
|
| 198 |
+
"learning_rate": 9.707046714172605e-06,
|
| 199 |
+
"loss": 0.2395,
|
| 200 |
+
"step": 1300
|
| 201 |
+
},
|
| 202 |
+
{
|
| 203 |
+
"epoch": 1.6033254156769596,
|
| 204 |
+
"grad_norm": 6.495739459991455,
|
| 205 |
+
"learning_rate": 9.311163895486937e-06,
|
| 206 |
+
"loss": 0.2266,
|
| 207 |
+
"step": 1350
|
| 208 |
+
},
|
| 209 |
+
{
|
| 210 |
+
"epoch": 1.66270783847981,
|
| 211 |
+
"grad_norm": 6.226797580718994,
|
| 212 |
+
"learning_rate": 8.915281076801267e-06,
|
| 213 |
+
"loss": 0.2179,
|
| 214 |
+
"step": 1400
|
| 215 |
+
},
|
| 216 |
+
{
|
| 217 |
+
"epoch": 1.7220902612826603,
|
| 218 |
+
"grad_norm": 5.022092819213867,
|
| 219 |
+
"learning_rate": 8.5193982581156e-06,
|
| 220 |
+
"loss": 0.2325,
|
| 221 |
+
"step": 1450
|
| 222 |
+
},
|
| 223 |
+
{
|
| 224 |
+
"epoch": 1.7814726840855108,
|
| 225 |
+
"grad_norm": 4.098015308380127,
|
| 226 |
+
"learning_rate": 8.12351543942993e-06,
|
| 227 |
+
"loss": 0.2307,
|
| 228 |
+
"step": 1500
|
| 229 |
+
},
|
| 230 |
+
{
|
| 231 |
+
"epoch": 1.8408551068883612,
|
| 232 |
+
"grad_norm": 5.266922473907471,
|
| 233 |
+
"learning_rate": 7.72763262074426e-06,
|
| 234 |
+
"loss": 0.2226,
|
| 235 |
+
"step": 1550
|
| 236 |
+
},
|
| 237 |
+
{
|
| 238 |
+
"epoch": 1.9002375296912115,
|
| 239 |
+
"grad_norm": 5.859071731567383,
|
| 240 |
+
"learning_rate": 7.331749802058591e-06,
|
| 241 |
+
"loss": 0.2251,
|
| 242 |
+
"step": 1600
|
| 243 |
+
},
|
| 244 |
+
{
|
| 245 |
+
"epoch": 1.9596199524940618,
|
| 246 |
+
"grad_norm": 4.122503757476807,
|
| 247 |
+
"learning_rate": 6.935866983372922e-06,
|
| 248 |
+
"loss": 0.2192,
|
| 249 |
+
"step": 1650
|
| 250 |
+
},
|
| 251 |
+
{
|
| 252 |
+
"epoch": 2.0,
|
| 253 |
+
"eval_accuracy": 0.8807339449541285,
|
| 254 |
+
"eval_loss": 0.36022087931632996,
|
| 255 |
+
"eval_runtime": 2.0552,
|
| 256 |
+
"eval_samples_per_second": 424.28,
|
| 257 |
+
"eval_steps_per_second": 2.919,
|
| 258 |
+
"step": 1684
|
| 259 |
+
},
|
| 260 |
+
{
|
| 261 |
+
"epoch": 2.019002375296912,
|
| 262 |
+
"grad_norm": 3.763842821121216,
|
| 263 |
+
"learning_rate": 6.539984164687253e-06,
|
| 264 |
+
"loss": 0.2036,
|
| 265 |
+
"step": 1700
|
| 266 |
+
},
|
| 267 |
+
{
|
| 268 |
+
"epoch": 2.0783847980997625,
|
| 269 |
+
"grad_norm": 5.909245014190674,
|
| 270 |
+
"learning_rate": 6.1441013460015846e-06,
|
| 271 |
+
"loss": 0.1885,
|
| 272 |
+
"step": 1750
|
| 273 |
+
},
|
| 274 |
+
{
|
| 275 |
+
"epoch": 2.137767220902613,
|
| 276 |
+
"grad_norm": 5.0692524909973145,
|
| 277 |
+
"learning_rate": 5.748218527315916e-06,
|
| 278 |
+
"loss": 0.1626,
|
| 279 |
+
"step": 1800
|
| 280 |
+
},
|
| 281 |
+
{
|
| 282 |
+
"epoch": 2.197149643705463,
|
| 283 |
+
"grad_norm": 6.9929633140563965,
|
| 284 |
+
"learning_rate": 5.352335708630247e-06,
|
| 285 |
+
"loss": 0.1844,
|
| 286 |
+
"step": 1850
|
| 287 |
+
},
|
| 288 |
+
{
|
| 289 |
+
"epoch": 2.2565320665083135,
|
| 290 |
+
"grad_norm": 6.867140769958496,
|
| 291 |
+
"learning_rate": 4.956452889944577e-06,
|
| 292 |
+
"loss": 0.1608,
|
| 293 |
+
"step": 1900
|
| 294 |
+
},
|
| 295 |
+
{
|
| 296 |
+
"epoch": 2.315914489311164,
|
| 297 |
+
"grad_norm": 7.411046981811523,
|
| 298 |
+
"learning_rate": 4.560570071258908e-06,
|
| 299 |
+
"loss": 0.1715,
|
| 300 |
+
"step": 1950
|
| 301 |
+
},
|
| 302 |
+
{
|
| 303 |
+
"epoch": 2.375296912114014,
|
| 304 |
+
"grad_norm": 5.924023151397705,
|
| 305 |
+
"learning_rate": 4.164687252573238e-06,
|
| 306 |
+
"loss": 0.1697,
|
| 307 |
+
"step": 2000
|
| 308 |
+
},
|
| 309 |
+
{
|
| 310 |
+
"epoch": 2.4346793349168645,
|
| 311 |
+
"grad_norm": 4.1883978843688965,
|
| 312 |
+
"learning_rate": 3.7688044338875695e-06,
|
| 313 |
+
"loss": 0.1687,
|
| 314 |
+
"step": 2050
|
| 315 |
+
},
|
| 316 |
+
{
|
| 317 |
+
"epoch": 2.494061757719715,
|
| 318 |
+
"grad_norm": 5.314006328582764,
|
| 319 |
+
"learning_rate": 3.3729216152019006e-06,
|
| 320 |
+
"loss": 0.1804,
|
| 321 |
+
"step": 2100
|
| 322 |
+
},
|
| 323 |
+
{
|
| 324 |
+
"epoch": 2.553444180522565,
|
| 325 |
+
"grad_norm": 6.460752487182617,
|
| 326 |
+
"learning_rate": 2.977038796516231e-06,
|
| 327 |
+
"loss": 0.1667,
|
| 328 |
+
"step": 2150
|
| 329 |
+
},
|
| 330 |
+
{
|
| 331 |
+
"epoch": 2.6128266033254155,
|
| 332 |
+
"grad_norm": 4.933835983276367,
|
| 333 |
+
"learning_rate": 2.581155977830562e-06,
|
| 334 |
+
"loss": 0.1717,
|
| 335 |
+
"step": 2200
|
| 336 |
+
},
|
| 337 |
+
{
|
| 338 |
+
"epoch": 2.6722090261282663,
|
| 339 |
+
"grad_norm": 4.656113624572754,
|
| 340 |
+
"learning_rate": 2.1852731591448932e-06,
|
| 341 |
+
"loss": 0.1703,
|
| 342 |
+
"step": 2250
|
| 343 |
+
},
|
| 344 |
+
{
|
| 345 |
+
"epoch": 2.731591448931116,
|
| 346 |
+
"grad_norm": 4.181103229522705,
|
| 347 |
+
"learning_rate": 1.789390340459224e-06,
|
| 348 |
+
"loss": 0.1887,
|
| 349 |
+
"step": 2300
|
| 350 |
+
},
|
| 351 |
+
{
|
| 352 |
+
"epoch": 2.790973871733967,
|
| 353 |
+
"grad_norm": 3.9885189533233643,
|
| 354 |
+
"learning_rate": 1.3935075217735553e-06,
|
| 355 |
+
"loss": 0.1735,
|
| 356 |
+
"step": 2350
|
| 357 |
+
},
|
| 358 |
+
{
|
| 359 |
+
"epoch": 2.850356294536817,
|
| 360 |
+
"grad_norm": 5.760321617126465,
|
| 361 |
+
"learning_rate": 9.976247030878861e-07,
|
| 362 |
+
"loss": 0.1721,
|
| 363 |
+
"step": 2400
|
| 364 |
+
},
|
| 365 |
+
{
|
| 366 |
+
"epoch": 2.9097387173396676,
|
| 367 |
+
"grad_norm": 4.699913501739502,
|
| 368 |
+
"learning_rate": 6.01741884402217e-07,
|
| 369 |
+
"loss": 0.1753,
|
| 370 |
+
"step": 2450
|
| 371 |
+
},
|
| 372 |
+
{
|
| 373 |
+
"epoch": 2.969121140142518,
|
| 374 |
+
"grad_norm": 6.697218894958496,
|
| 375 |
+
"learning_rate": 2.0585906571654793e-07,
|
| 376 |
+
"loss": 0.1742,
|
| 377 |
+
"step": 2500
|
| 378 |
+
},
|
| 379 |
+
{
|
| 380 |
+
"epoch": 3.0,
|
| 381 |
+
"eval_accuracy": 0.8727064220183486,
|
| 382 |
+
"eval_loss": 0.3852188289165497,
|
| 383 |
+
"eval_runtime": 2.2933,
|
| 384 |
+
"eval_samples_per_second": 380.244,
|
| 385 |
+
"eval_steps_per_second": 2.616,
|
| 386 |
+
"step": 2526
|
| 387 |
+
},
|
| 388 |
+
{
|
| 389 |
+
"epoch": 3.0,
|
| 390 |
+
"step": 2526,
|
| 391 |
+
"total_flos": 6633975382115328.0,
|
| 392 |
+
"train_loss": 0.25833009861804906,
|
| 393 |
+
"train_runtime": 268.2911,
|
| 394 |
+
"train_samples_per_second": 753.089,
|
| 395 |
+
"train_steps_per_second": 9.415
|
| 396 |
+
}
|
| 397 |
+
],
|
| 398 |
+
"logging_steps": 50,
|
| 399 |
+
"max_steps": 2526,
|
| 400 |
+
"num_input_tokens_seen": 0,
|
| 401 |
+
"num_train_epochs": 3,
|
| 402 |
+
"save_steps": 500,
|
| 403 |
+
"stateful_callbacks": {
|
| 404 |
+
"TrainerControl": {
|
| 405 |
+
"args": {
|
| 406 |
+
"should_epoch_stop": false,
|
| 407 |
+
"should_evaluate": false,
|
| 408 |
+
"should_log": false,
|
| 409 |
+
"should_save": false,
|
| 410 |
+
"should_training_stop": false
|
| 411 |
+
},
|
| 412 |
+
"attributes": {}
|
| 413 |
+
}
|
| 414 |
+
},
|
| 415 |
+
"total_flos": 6633975382115328.0,
|
| 416 |
+
"train_batch_size": 80,
|
| 417 |
+
"trial_name": null,
|
| 418 |
+
"trial_params": null
|
| 419 |
+
}
|