Training in progress, step 50

all_results.json

@@ -1,18 +1,18 @@
 {
-    "epoch": 20.0,
-    "eval_accuracy": 57.99999999999999,
-    "eval_average_metrics": 57.99999999999999,
-    "eval_loss": 0.2818337678909302,
-    "eval_runtime": 0.316,
-    "eval_samples_per_second": 158.207,
-    "test_accuracy": 54.0,
-    "test_average_metrics": 54.0,
-    "test_loss": 0.2671838104724884,
-    "test_runtime": 0.2712,
-    "test_samples_per_second": 184.354,
-    "train_loss": 0.34575576782226564,
-    "train_runtime": 106.0291,
+    "epoch": 40.0,
+    "eval_accuracy": 54.0,
+    "eval_average_metrics": 54.0,
+    "eval_loss": 0.29293137788772583,
+    "eval_runtime": 0.3662,
+    "eval_samples_per_second": 136.552,
+    "test_accuracy": 52.0,
+    "test_average_metrics": 52.0,
+    "test_loss": 0.28209835290908813,
+    "test_runtime": 0.3997,
+    "test_samples_per_second": 125.094,
+    "train_loss": 0.18321549898156753,
+    "train_runtime": 219.7861,
     "train_samples": 400,
-    "train_samples_per_second": 75.451,
-    "train_steps_per_second": 2.452
+    "train_samples_per_second": 72.798,
+    "train_steps_per_second": 2.366
 }

eval_results.json

@@ -1,8 +1,8 @@
 {
-    "epoch": 20.0,
-    "eval_accuracy": 57.99999999999999,
-    "eval_average_metrics": 57.99999999999999,
-    "eval_loss": 0.2818337678909302,
-    "eval_runtime": 0.316,
-    "eval_samples_per_second": 158.207
+    "epoch": 40.0,
+    "eval_accuracy": 54.0,
+    "eval_average_metrics": 54.0,
+    "eval_loss": 0.29293137788772583,
+    "eval_runtime": 0.3662,
+    "eval_samples_per_second": 136.552
 }

pytorch_model.bin

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:b2709d3778630b2d21b2604feac69f244a13e550bf3c664aefd8ea3aeacc9789
-size 7551621
+oid sha256:615d83a030dfaea92497a8df95a41acb7f6517ebddd6b1b87e1356174aaad9b1
+size 2631685

runs/Feb01_01-06-37_node1/events.out.tfevents.1643648950.node1

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:93e86d1962aa4528dbdfe41a02d0aef59a7c2fa5e1b86383ff4d57f1c1dd87b6
-size 7371
+oid sha256:f755d9d8522438f177d99db906ad50f4e49819757e581a89654b3e23b4dd2975
+size 8047

runs/Feb01_01-06-37_node1/events.out.tfevents.1643649170.node1

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:c145b61bc84703439a6304ebe14d0ce65c3d44fd6dfce886724d9823c76198ac
+size 684

runs/Feb02_15-36-39_node1/1643787501.4051788/events.out.tfevents.1643787501.node1

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:45de942c7728aa9c2815f1f011924005de5494c578a6c8a31ef2c9594c94b3af
+size 5041

runs/Feb02_15-36-39_node1/events.out.tfevents.1643787501.node1

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:b2ef62b884b6d099805f3ce90522786e70e966c8880b23400afa20e402af4fed
+size 4328

test_results.json

@@ -1,8 +1,8 @@
 {
-    "epoch": 20.0,
-    "test_accuracy": 54.0,
-    "test_average_metrics": 54.0,
-    "test_loss": 0.2671838104724884,
-    "test_runtime": 0.2712,
-    "test_samples_per_second": 184.354
+    "epoch": 40.0,
+    "test_accuracy": 52.0,
+    "test_average_metrics": 52.0,
+    "test_loss": 0.28209835290908813,
+    "test_runtime": 0.3997,
+    "test_samples_per_second": 125.094
 }

train_results.json

@@ -1,8 +1,8 @@
 {
-    "epoch": 20.0,
-    "train_loss": 0.34575576782226564,
-    "train_runtime": 106.0291,
+    "epoch": 40.0,
+    "train_loss": 0.18321549898156753,
+    "train_runtime": 219.7861,
     "train_samples": 400,
-    "train_samples_per_second": 75.451,
-    "train_steps_per_second": 2.452
+    "train_samples_per_second": 72.798,
+    "train_steps_per_second": 2.366
 }

trainer_state.json

@@ -1,70 +1,121 @@
 {
-  "best_metric": 57.99999999999999,
-  "best_model_checkpoint": "outputs/bitfit/t5-base/superglue-copa/checkpoint-250",
-  "epoch": 20.0,
-  "global_step": 260,
+  "best_metric": 54.0,
+  "best_model_checkpoint": "outputs/bitfit/t5-base/superglue-copa/checkpoint-200",
+  "epoch": 40.0,
+  "global_step": 520,
   "is_hyper_param_search": false,
   "is_local_process_zero": true,
   "is_world_process_zero": true,
   "log_history": [
     {
       "epoch": 3.85,
-      "eval_accuracy": 48.0,
-      "eval_average_metrics": 48.0,
-      "eval_loss": 0.2776777148246765,
-      "eval_runtime": 0.5637,
-      "eval_samples_per_second": 88.704,
+      "eval_accuracy": 50.0,
+      "eval_average_metrics": 50.0,
+      "eval_loss": 0.2800115942955017,
+      "eval_runtime": 0.3692,
+      "eval_samples_per_second": 135.418,
       "step": 50
     },
     {
       "epoch": 7.69,
-      "eval_accuracy": 50.0,
-      "eval_average_metrics": 50.0,
-      "eval_loss": 0.27705639600753784,
-      "eval_runtime": 0.2608,
-      "eval_samples_per_second": 191.742,
+      "eval_accuracy": 42.0,
+      "eval_average_metrics": 42.0,
+      "eval_loss": 0.28102436661720276,
+      "eval_runtime": 0.4452,
+      "eval_samples_per_second": 112.304,
       "step": 100
     },
     {
       "epoch": 11.54,
-      "eval_accuracy": 48.0,
-      "eval_average_metrics": 48.0,
-      "eval_loss": 0.2773664593696594,
-      "eval_runtime": 0.3349,
-      "eval_samples_per_second": 149.291,
+      "eval_accuracy": 52.0,
+      "eval_average_metrics": 52.0,
+      "eval_loss": 0.3020976185798645,
+      "eval_runtime": 0.3114,
+      "eval_samples_per_second": 160.544,
       "step": 150
     },
     {
       "epoch": 15.38,
-      "eval_accuracy": 44.0,
-      "eval_average_metrics": 44.0,
-      "eval_loss": 0.28644633293151855,
-      "eval_runtime": 0.2612,
-      "eval_samples_per_second": 191.401,
+      "eval_accuracy": 54.0,
+      "eval_average_metrics": 54.0,
+      "eval_loss": 0.29293137788772583,
+      "eval_runtime": 0.5099,
+      "eval_samples_per_second": 98.067,
       "step": 200
     },
     {
       "epoch": 19.23,
-      "eval_accuracy": 57.99999999999999,
-      "eval_average_metrics": 57.99999999999999,
-      "eval_loss": 0.2818337678909302,
-      "eval_runtime": 0.2709,
-      "eval_samples_per_second": 184.56,
+      "eval_accuracy": 50.0,
+      "eval_average_metrics": 50.0,
+      "eval_loss": 0.4718337953090668,
+      "eval_runtime": 0.2966,
+      "eval_samples_per_second": 168.586,
       "step": 250
     },
     {
-      "epoch": 20.0,
-      "step": 260,
-      "total_flos": 477340229431296.0,
-      "train_loss": 0.34575576782226564,
-      "train_runtime": 106.0291,
-      "train_samples_per_second": 75.451,
-      "train_steps_per_second": 2.452
+      "epoch": 23.08,
+      "eval_accuracy": 48.0,
+      "eval_average_metrics": 48.0,
+      "eval_loss": 0.6855034828186035,
+      "eval_runtime": 0.3659,
+      "eval_samples_per_second": 136.664,
+      "step": 300
+    },
+    {
+      "epoch": 26.92,
+      "eval_accuracy": 48.0,
+      "eval_average_metrics": 48.0,
+      "eval_loss": 0.9259137511253357,
+      "eval_runtime": 0.5574,
+      "eval_samples_per_second": 89.705,
+      "step": 350
+    },
+    {
+      "epoch": 30.77,
+      "eval_accuracy": 44.0,
+      "eval_average_metrics": 44.0,
+      "eval_loss": 1.0988849401474,
+      "eval_runtime": 0.3043,
+      "eval_samples_per_second": 164.333,
+      "step": 400
+    },
+    {
+      "epoch": 34.62,
+      "eval_accuracy": 50.0,
+      "eval_average_metrics": 50.0,
+      "eval_loss": 1.1635206937789917,
+      "eval_runtime": 0.3906,
+      "eval_samples_per_second": 128.015,
+      "step": 450
+    },
+    {
+      "epoch": 38.46,
+      "learning_rate": 1.1538461538461538e-05,
+      "loss": 0.19,
+      "step": 500
+    },
+    {
+      "epoch": 38.46,
+      "eval_accuracy": 48.0,
+      "eval_average_metrics": 48.0,
+      "eval_loss": 1.2427139282226562,
+      "eval_runtime": 0.3254,
+      "eval_samples_per_second": 153.656,
+      "step": 500
+    },
+    {
+      "epoch": 40.0,
+      "step": 520,
+      "total_flos": 954277186351104.0,
+      "train_loss": 0.18321549898156753,
+      "train_runtime": 219.7861,
+      "train_samples_per_second": 72.798,
+      "train_steps_per_second": 2.366
     }
   ],
-  "max_steps": 260,
-  "num_train_epochs": 20,
-  "total_flos": 477340229431296.0,
+  "max_steps": 520,
+  "num_train_epochs": 40,
+  "total_flos": 954277186351104.0,
   "trial_name": null,
   "trial_params": null
 }

training_args.bin

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:1eb01bad7c8b0307ff6935ee0dbe2d8437e8b5d224ca89a00e9120095551b79d
+oid sha256:cd7c9dde74bac513e1fc0b66b0b2e4cc06aea4079142ec3c5f7d32d3e1397850
 size 3183

training_config.json

@@ -1 +1 @@
-{"bottleneck_dim": 24, "dataset_config_name": ["en"], "delta_type": "adapter", "do_eval": true, "do_test": true, "do_train": true, "eval_dataset_config_name": ["en"], "eval_dataset_name": "superglue-copa", "eval_steps": 50, "evaluation_strategy": "steps", "greater_is_better": true, "learning_rate": 0.0003, "load_best_model_at_end": true, "max_source_length": 256, "metric_for_best_model": "average_metrics", "model_name_or_path": "../../../../plm_cache/t5-base", "num_train_epochs": 40, "output_dir": "outputs/bitfit/t5-base/superglue-copa", "overwrite_output_dir": true, "per_device_eval_batch_size": 32, "per_device_train_batch_size": 32, "predict_with_generate": true, "push_to_hub": true, "save_steps": 50, "save_strategy": "steps", "save_total_limit": 1, "seed": 42, "split_validation_test": true, "task_name": "superglue-copa", "test_dataset_config_name": ["en"], "test_dataset_name": "superglue-copa", "tokenizer_name": "../../../../plm_cache/t5-base", "unfrozen_modules": ["deltas", "layer_norm", "final_layer_norm"], "warmup_steps": 0}
+{"dataset_config_name": ["en"], "delta_type": "lora", "do_eval": true, "do_test": true, "do_train": true, "eval_dataset_config_name": ["en"], "eval_dataset_name": "superglue-copa", "eval_steps": 50, "evaluation_strategy": "steps", "greater_is_better": true, "learning_rate": 0.0003, "load_best_model_at_end": true, "lora_r": 8, "max_source_length": 256, "metric_for_best_model": "average_metrics", "model_name_or_path": "../../../../plm_cache/t5-base", "num_train_epochs": 40, "output_dir": "outputs/bitfit/t5-base/superglue-copa", "overwrite_output_dir": true, "per_device_eval_batch_size": 32, "per_device_train_batch_size": 32, "predict_with_generate": true, "push_to_hub": true, "save_steps": 50, "save_strategy": "steps", "save_total_limit": 1, "seed": 42, "split_validation_test": true, "task_name": "superglue-copa", "test_dataset_config_name": ["en"], "test_dataset_name": "superglue-copa", "tokenizer_name": "../../../../plm_cache/t5-base", "unfrozen_modules": ["deltas", "layer_norm", "final_layer_norm"], "warmup_steps": 0}