Training in progress, step 50

all_results.json

@@ -1,18 +1,18 @@
 {
     "epoch": 20.0,
-    "eval_accuracy": 44.0,
-    "eval_average_metrics": 44.0,
-    "eval_loss": 0.28644633293151855,
-    "eval_runtime": 0.3326,
-    "eval_samples_per_second": 150.31,
-    "test_accuracy": 44.0,
-    "test_average_metrics": 44.0,
-    "test_loss": 0.2822103500366211,
-    "test_runtime": 0.5342,
-    "test_samples_per_second": 93.601,
+    "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": 70.261,
+    "train_runtime": 106.0291,
     "train_samples": 400,
-    "train_samples_per_second": 113.861,
-    "train_steps_per_second": 3.7
+    "train_samples_per_second": 75.451,
+    "train_steps_per_second": 2.452
 }

eval_results.json

@@ -1,8 +1,8 @@
 {
     "epoch": 20.0,
-    "eval_accuracy": 44.0,
-    "eval_average_metrics": 44.0,
-    "eval_loss": 0.28644633293151855,
-    "eval_runtime": 0.3326,
-    "eval_samples_per_second": 150.31
+    "eval_accuracy": 57.99999999999999,
+    "eval_average_metrics": 57.99999999999999,
+    "eval_loss": 0.2818337678909302,
+    "eval_runtime": 0.316,
+    "eval_samples_per_second": 158.207
 }

pytorch_model.bin

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:8a1e5021796764be1ac2c8dc1ab7d2e720ac35706fe360d3451363c8252f8d7d
+oid sha256:e1a143b9a37d7c4a9e6b90e95a66dfe8bb8a5db5c678450da91efb102c44e93c
 size 7551621

runs/Feb01_00-55-21_node1/events.out.tfevents.1643648238.node1

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:987bf59dc742210869f8e64eabd6037177b23965bf0e80e8cd4e5667a7821d90
-size 5604
+oid sha256:20702b71ecddb28b4d1774d9709dccc50c873d832b045f60a67f11c77726bf77
+size 6280

runs/Feb01_00-55-21_node1/events.out.tfevents.1643648344.node1

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:9a878c42713e7ed4cf3578987cb5f1b481a2592a68a95f0ac94dd4ad2eea8933
+size 684

runs/Feb01_01-06-37_node1/1643648950.4255028/events.out.tfevents.1643648950.node1

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:af90228ddce35b7747c02f23e3aafc3bfef4c17a92f9f56ab3c90da436f20743
+size 5041

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

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:f9bb3626b905ccf6f2eba85bdc6e7be71c18357e09025cd990a2ce5a79c1988a
+size 4328

test_results.json

@@ -1,8 +1,8 @@
 {
     "epoch": 20.0,
-    "test_accuracy": 44.0,
-    "test_average_metrics": 44.0,
-    "test_loss": 0.2822103500366211,
-    "test_runtime": 0.5342,
-    "test_samples_per_second": 93.601
+    "test_accuracy": 54.0,
+    "test_average_metrics": 54.0,
+    "test_loss": 0.2671838104724884,
+    "test_runtime": 0.2712,
+    "test_samples_per_second": 184.354
 }

train_results.json

@@ -1,8 +1,8 @@
 {
     "epoch": 20.0,
     "train_loss": 0.34575576782226564,
-    "train_runtime": 70.261,
+    "train_runtime": 106.0291,
     "train_samples": 400,
-    "train_samples_per_second": 113.861,
-    "train_steps_per_second": 3.7
+    "train_samples_per_second": 75.451,
+    "train_steps_per_second": 2.452
 }

trainer_state.json

@@ -1,29 +1,65 @@
 {
-  "best_metric": 44.0,
-  "best_model_checkpoint": "outputs/bitfit/t5-base/superglue-copa/checkpoint-200",
+  "best_metric": 57.99999999999999,
+  "best_model_checkpoint": "outputs/bitfit/t5-base/superglue-copa/checkpoint-250",
   "epoch": 20.0,
   "global_step": 260,
   "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,
+      "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,
+      "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,
+      "step": 150
+    },
     {
       "epoch": 15.38,
       "eval_accuracy": 44.0,
       "eval_average_metrics": 44.0,
       "eval_loss": 0.28644633293151855,
-      "eval_runtime": 0.4561,
-      "eval_samples_per_second": 109.613,
+      "eval_runtime": 0.2612,
+      "eval_samples_per_second": 191.401,
       "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,
+      "step": 250
+    },
     {
       "epoch": 20.0,
       "step": 260,
       "total_flos": 477340229431296.0,
       "train_loss": 0.34575576782226564,
-      "train_runtime": 70.261,
-      "train_samples_per_second": 113.861,
-      "train_steps_per_second": 3.7
+      "train_runtime": 106.0291,
+      "train_samples_per_second": 75.451,
+      "train_steps_per_second": 2.452
     }
   ],
   "max_steps": 260,

training_args.bin

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:03b7f77ffc2d3919af0bd693d9443ed1c8b8f660e4a128129e827db5910934c4
+oid sha256:1eb01bad7c8b0307ff6935ee0dbe2d8437e8b5d224ca89a00e9120095551b79d
 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": 20, "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}
+{"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}