Training in progress, step 100

all_results.json

@@ -1,18 +1,18 @@
 {
     "epoch": 20.0,
-    "eval_accuracy": 55.769230769230774,
-    "eval_average_metrics": 55.769230769230774,
-    "eval_loss": 0.26899516582489014,
-    "eval_runtime": 0.2241,
-    "eval_samples_per_second": 232.063,
-    "test_accuracy": 61.53846153846154,
-    "test_average_metrics": 61.53846153846154,
-    "test_loss": 0.2528702914714813,
-    "test_runtime": 0.2148,
-    "test_samples_per_second": 242.088,
-    "train_loss": 0.4271513197157118,
-    "train_runtime": 68.8963,
+    "eval_accuracy": 59.61538461538461,
+    "eval_average_metrics": 59.61538461538461,
+    "eval_loss": 0.2665896713733673,
+    "eval_runtime": 0.3622,
+    "eval_samples_per_second": 143.571,
+    "test_accuracy": 67.3076923076923,
+    "test_average_metrics": 67.3076923076923,
+    "test_loss": 0.25851768255233765,
+    "test_runtime": 0.3561,
+    "test_samples_per_second": 146.039,
+    "train_loss": 0.33580604129367403,
+    "train_runtime": 150.2227,
     "train_samples": 554,
-    "train_samples_per_second": 160.821,
-    "train_steps_per_second": 5.225
+    "train_samples_per_second": 73.757,
+    "train_steps_per_second": 2.396
 }

eval_results.json

@@ -1,8 +1,8 @@
 {
     "epoch": 20.0,
-    "eval_accuracy": 55.769230769230774,
-    "eval_average_metrics": 55.769230769230774,
-    "eval_loss": 0.26899516582489014,
-    "eval_runtime": 0.2241,
-    "eval_samples_per_second": 232.063
+    "eval_accuracy": 59.61538461538461,
+    "eval_average_metrics": 59.61538461538461,
+    "eval_loss": 0.2665896713733673,
+    "eval_runtime": 0.3622,
+    "eval_samples_per_second": 143.571
 }

pytorch_model.bin

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:3068e961e1a25769ce6b5446bb76d5e472ddea9d678b095aec446aff796cb6ea
-size 7551621
+oid sha256:cf7bc90659290656cca9a44b62c7ad84266260470b3ed8c50ebca550d91e0dc7
+size 2631685

runs/Feb01_07-17-34_node1/events.out.tfevents.1643671165.node1

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:92cd82800afd27d7b811a85caf1931d30aec696377833625c5b5ad2f20d1d9d2
-size 5305
+oid sha256:d4ce4600c32ce7d25a503e8e9c844e49f75fd783f9978948a57c60fc3ce66f2e
+size 5659

runs/Feb01_07-17-34_node1/events.out.tfevents.1643671316.node1

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:8c994c838e84902c90c9521a2b776f6a12331a7a3d000613d3119cb6a2c6ec47
+size 684

runs/Feb02_18-32-00_node1/1643798001.1914194/events.out.tfevents.1643798001.node1

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:eee15a7a0afd6ebed8417a925aefc56d1642daa62be706bec86f386f847c1f6a
+size 5056

runs/Feb02_18-32-00_node1/events.out.tfevents.1643798001.node1

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:eaa36c24fe9d70c2cbcb4f751264a5690685f05ca7cec0d0b0987e6d655cb273
+size 4345

test_results.json

@@ -1,8 +1,8 @@
 {
     "epoch": 20.0,
-    "test_accuracy": 61.53846153846154,
-    "test_average_metrics": 61.53846153846154,
-    "test_loss": 0.2528702914714813,
-    "test_runtime": 0.2148,
-    "test_samples_per_second": 242.088
+    "test_accuracy": 67.3076923076923,
+    "test_average_metrics": 67.3076923076923,
+    "test_loss": 0.25851768255233765,
+    "test_runtime": 0.3561,
+    "test_samples_per_second": 146.039
 }

train_results.json

@@ -1,8 +1,8 @@
 {
     "epoch": 20.0,
-    "train_loss": 0.4271513197157118,
-    "train_runtime": 68.8963,
+    "train_loss": 0.33580604129367403,
+    "train_runtime": 150.2227,
     "train_samples": 554,
-    "train_samples_per_second": 160.821,
-    "train_steps_per_second": 5.225
+    "train_samples_per_second": 73.757,
+    "train_steps_per_second": 2.396
 }

trainer_state.json

@@ -1,34 +1,52 @@
 {
-  "best_metric": 55.769230769230774,
-  "best_model_checkpoint": "outputs/bitfit/t5-base/superglue-wsc.fixed/checkpoint-200",
+  "best_metric": 59.61538461538461,
+  "best_model_checkpoint": "outputs/bitfit/t5-base/superglue-wsc.fixed/checkpoint-300",
   "epoch": 20.0,
   "global_step": 360,
   "is_hyper_param_search": false,
   "is_local_process_zero": true,
   "is_world_process_zero": true,
   "log_history": [
+    {
+      "epoch": 5.56,
+      "eval_accuracy": 40.38461538461539,
+      "eval_average_metrics": 40.38461538461539,
+      "eval_loss": 0.33406707644462585,
+      "eval_runtime": 0.2912,
+      "eval_samples_per_second": 178.549,
+      "step": 100
+    },
     {
       "epoch": 11.11,
-      "eval_accuracy": 55.769230769230774,
-      "eval_average_metrics": 55.769230769230774,
-      "eval_loss": 0.26899516582489014,
-      "eval_runtime": 0.4189,
-      "eval_samples_per_second": 124.126,
+      "eval_accuracy": 40.38461538461539,
+      "eval_average_metrics": 40.38461538461539,
+      "eval_loss": 0.27627798914909363,
+      "eval_runtime": 0.3043,
+      "eval_samples_per_second": 170.86,
       "step": 200
     },
+    {
+      "epoch": 16.67,
+      "eval_accuracy": 59.61538461538461,
+      "eval_average_metrics": 59.61538461538461,
+      "eval_loss": 0.2665896713733673,
+      "eval_runtime": 0.2321,
+      "eval_samples_per_second": 224.02,
+      "step": 300
+    },
     {
       "epoch": 20.0,
       "step": 360,
-      "total_flos": 1490349041823024.0,
-      "train_loss": 0.4271513197157118,
-      "train_runtime": 68.8963,
-      "train_samples_per_second": 160.821,
-      "train_steps_per_second": 5.225
+      "total_flos": 1502077285006848.0,
+      "train_loss": 0.33580604129367403,
+      "train_runtime": 150.2227,
+      "train_samples_per_second": 73.757,
+      "train_steps_per_second": 2.396
     }
   ],
   "max_steps": 360,
   "num_train_epochs": 20,
-  "total_flos": 1490349041823024.0,
+  "total_flos": 1502077285006848.0,
   "trial_name": null,
   "trial_params": null
 }

training_args.bin

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:7bd02b28bab24c1face8a1dc29c40509cef5a1f6da4574d8ba3a16d8da091037
+oid sha256:d63e3cd41ee78796a83c6d07d897953f0dc58f62def755e00f36e1aa2e030b96
 size 3247

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-wsc.fixed", "eval_steps": 100, "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-wsc.fixed", "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": 100, "save_strategy": "steps", "save_total_limit": 1, "seed": 42, "split_validation_test": true, "task_name": "superglue-wsc.fixed", "test_dataset_config_name": ["en"], "test_dataset_name": "superglue-wsc.fixed", "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-wsc.fixed", "eval_steps": 100, "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": 20, "output_dir": "outputs/bitfit/t5-base/superglue-wsc.fixed", "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": 100, "save_strategy": "steps", "save_total_limit": 1, "seed": 42, "split_validation_test": true, "task_name": "superglue-wsc.fixed", "test_dataset_config_name": ["en"], "test_dataset_name": "superglue-wsc.fixed", "tokenizer_name": "../../../../plm_cache/t5-base", "unfrozen_modules": ["deltas", "layer_norm", "final_layer_norm"], "warmup_steps": 0}