Training in progress, step 50

all_results.json

@@ -1,18 +1,18 @@
 {
     "epoch": 20.0,
-    "eval_accuracy": 40.0,
-    "eval_average_metrics": 40.0,
-    "eval_loss": 0.2866286337375641,
-    "eval_runtime": 0.149,
-    "eval_samples_per_second": 335.514,
+    "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": 0.2718357443809509,
-    "test_runtime": 0.1482,
-    "test_samples_per_second": 337.428,
-    "train_loss": 0.47191429138183594,
-    "train_runtime": 30.868,
+    "test_loss": 0.2822103500366211,
+    "test_runtime": 0.5342,
+    "test_samples_per_second": 93.601,
+    "train_loss": 0.34575576782226564,
+    "train_runtime": 70.261,
     "train_samples": 400,
-    "train_samples_per_second": 259.168,
-    "train_steps_per_second": 8.423
+    "train_samples_per_second": 113.861,
+    "train_steps_per_second": 3.7
 }

eval_results.json

@@ -1,8 +1,8 @@
 {
     "epoch": 20.0,
-    "eval_accuracy": 40.0,
-    "eval_average_metrics": 40.0,
-    "eval_loss": 0.2866286337375641,
-    "eval_runtime": 0.149,
-    "eval_samples_per_second": 335.514
+    "eval_accuracy": 44.0,
+    "eval_average_metrics": 44.0,
+    "eval_loss": 0.28644633293151855,
+    "eval_runtime": 0.3326,
+    "eval_samples_per_second": 150.31
 }

pytorch_model.bin

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:c45abc26f5ac48d57c9e9ef202ed9e87c6a866c812ad30088af46f5f53437668
+oid sha256:016e16eafff8f959d9056a2bdd5a29cf07d59210b35b36c12b61b15d08787d64
 size 7551621

runs/Feb01_00-38-26_node1/events.out.tfevents.1643647198.node1

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:84c39b517335eb334fbd10f035f257954f57adad0b8b77984e93bad4b534b9a5
-size 4330
+oid sha256:1ad7d7cf7e820ced97e901d97c39d3d4b68f9b82d95445a9a919a36c01f65998
+size 5006

runs/Feb01_00-38-26_node1/events.out.tfevents.1643647269.node1

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:2224ae6dc6ddd0224a3e4a4fb10367c12cd6303cbb7e2860fa789a77c5333df1
+size 684

runs/Feb01_00-55-21_node1/1643648238.0535004/events.out.tfevents.1643648238.node1

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:61337dea3b2d4f940d00e63f4d9fba853734c0a2ce41d31c4dd9070aa5d4dd97
+size 5041

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

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:57683a0dfa1546984cf394cd93b4c4da9cc3eac45975ce98dc9f665ea4ff3aca
+size 4328

test_results.json

@@ -2,7 +2,7 @@
     "epoch": 20.0,
     "test_accuracy": 44.0,
     "test_average_metrics": 44.0,
-    "test_loss": 0.2718357443809509,
-    "test_runtime": 0.1482,
-    "test_samples_per_second": 337.428
+    "test_loss": 0.2822103500366211,
+    "test_runtime": 0.5342,
+    "test_samples_per_second": 93.601
 }

train_results.json

@@ -1,8 +1,8 @@
 {
     "epoch": 20.0,
-    "train_loss": 0.47191429138183594,
-    "train_runtime": 30.868,
+    "train_loss": 0.34575576782226564,
+    "train_runtime": 70.261,
     "train_samples": 400,
-    "train_samples_per_second": 259.168,
-    "train_steps_per_second": 8.423
+    "train_samples_per_second": 113.861,
+    "train_steps_per_second": 3.7
 }

trainer_state.json

@@ -1,5 +1,5 @@
 {
-  "best_metric": 40.0,
+  "best_metric": 44.0,
   "best_model_checkpoint": "outputs/bitfit/t5-base/superglue-copa/checkpoint-200",
   "epoch": 20.0,
   "global_step": 260,
@@ -9,26 +9,26 @@
   "log_history": [
     {
       "epoch": 15.38,
-      "eval_accuracy": 40.0,
-      "eval_average_metrics": 40.0,
-      "eval_loss": 0.2866286337375641,
-      "eval_runtime": 0.1965,
-      "eval_samples_per_second": 254.502,
+      "eval_accuracy": 44.0,
+      "eval_average_metrics": 44.0,
+      "eval_loss": 0.28644633293151855,
+      "eval_runtime": 0.4561,
+      "eval_samples_per_second": 109.613,
       "step": 200
     },
     {
       "epoch": 20.0,
       "step": 260,
-      "total_flos": 473613149374848.0,
-      "train_loss": 0.47191429138183594,
-      "train_runtime": 30.868,
-      "train_samples_per_second": 259.168,
-      "train_steps_per_second": 8.423
+      "total_flos": 477340229431296.0,
+      "train_loss": 0.34575576782226564,
+      "train_runtime": 70.261,
+      "train_samples_per_second": 113.861,
+      "train_steps_per_second": 3.7
     }
   ],
   "max_steps": 260,
   "num_train_epochs": 20,
-  "total_flos": 473613149374848.0,
+  "total_flos": 477340229431296.0,
   "trial_name": null,
   "trial_params": null
 }

training_args.bin

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:f3cd333362728449f5def839245c02b44ab8344c98cb431a1adedfccbd55dcce
+oid sha256:03b7f77ffc2d3919af0bd693d9443ed1c8b8f660e4a128129e827db5910934c4
 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": 200, "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": 200, "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": 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}