Training in progress, step 100

all_results.json

@@ -3,18 +3,18 @@
     "eval_accuracy": 81.15942028985508,
     "eval_average_metrics": 81.15942028985508,
     "eval_loss": 0.18789426982402802,
-    "eval_runtime": 0.4156,
+    "eval_runtime": 0.6932,
     "eval_samples": 277,
-    "eval_samples_per_second": 332.064,
+    "eval_samples_per_second": 199.08,
     "eval_steps_per_second": 6.498,
     "test_accuracy": 69.7841726618705,
     "test_average_metrics": 69.7841726618705,
     "test_loss": 0.23357094824314117,
-    "test_runtime": 0.5204,
-    "test_samples_per_second": 267.078,
+    "test_runtime": 0.4076,
+    "test_samples_per_second": 340.999,
     "train_loss": 0.7619044189453125,
-    "train_runtime": 190.0328,
+    "train_runtime": 235.5565,
     "train_samples": 2490,
-    "train_samples_per_second": 262.06,
-    "train_steps_per_second": 2.631
+    "train_samples_per_second": 211.414,
+    "train_steps_per_second": 2.123
 }

eval_results.json

@@ -3,6 +3,6 @@
     "eval_accuracy": 81.15942028985508,
     "eval_average_metrics": 81.15942028985508,
     "eval_loss": 0.18789426982402802,
-    "eval_runtime": 0.4156,
-    "eval_samples_per_second": 332.064
+    "eval_runtime": 0.6932,
+    "eval_samples_per_second": 199.08
 }

pytorch_model.bin

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:5064b631760eb53b6d4411fa4fad080087100c3c7d5afa39f4ab7c694c90449f
+oid sha256:6c5f0d8e184126d8020f2064f690cf09dffc2804d04409edf663d8ecd56718a6
 size 7551621

runs/Feb01_00-55-50_node2/events.out.tfevents.1643648239.node2

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:4e9577db74a4cb0a700b8dd3ba82ed4d507bd0fd4a37a9416cac27d0940de909
-size 5740
+oid sha256:a485fb9270b0811b824ccf6c6f7697b53a5f915a41026acf3bd9b4a5d542cb92
+size 6416

runs/Feb01_00-55-50_node2/events.out.tfevents.1643648476.node2

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:d12533e9322e7aab4ac1030379d4b0ffdb90e70b4bedbcf0ab5f34694a0b6138
+size 684

runs/Feb01_01-06-20_node2/1643648862.5479631/events.out.tfevents.1643648862.node2

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:68bb913d806baf8b6c746d64fb5131cf0ba29fb2ba73713d384942beda977bb7
+size 5008

runs/Feb01_01-06-20_node2/events.out.tfevents.1643648862.node2

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:c84d62476df9daf5137047611167728d0bf730cdae325819c211675b58a767ab
+size 4297

test_results.json

@@ -3,6 +3,6 @@
     "test_accuracy": 69.7841726618705,
     "test_average_metrics": 69.7841726618705,
     "test_loss": 0.23357094824314117,
-    "test_runtime": 0.5204,
-    "test_samples_per_second": 267.078
+    "test_runtime": 0.4076,
+    "test_samples_per_second": 340.999
 }

train_results.json

@@ -1,8 +1,8 @@
 {
     "epoch": 20.0,
     "train_loss": 0.7619044189453125,
-    "train_runtime": 190.0328,
+    "train_runtime": 235.5565,
     "train_samples": 2490,
-    "train_samples_per_second": 262.06,
-    "train_steps_per_second": 2.631
+    "train_samples_per_second": 211.414,
+    "train_steps_per_second": 2.123
 }

trainer_state.json

@@ -7,22 +7,40 @@
   "is_local_process_zero": true,
   "is_world_process_zero": true,
   "log_history": [
+    {
+      "epoch": 4.0,
+      "eval_accuracy": 55.79710144927537,
+      "eval_average_metrics": 55.79710144927537,
+      "eval_loss": 0.2813531160354614,
+      "eval_runtime": 0.6184,
+      "eval_samples_per_second": 223.144,
+      "step": 100
+    },
     {
       "epoch": 8.0,
       "eval_accuracy": 81.15942028985508,
       "eval_average_metrics": 81.15942028985508,
       "eval_loss": 0.18789426982402802,
-      "eval_runtime": 0.5472,
-      "eval_samples_per_second": 252.185,
+      "eval_runtime": 0.5049,
+      "eval_samples_per_second": 273.329,
       "step": 200
     },
+    {
+      "epoch": 12.0,
+      "eval_accuracy": 76.81159420289855,
+      "eval_average_metrics": 76.81159420289855,
+      "eval_loss": 0.18861950933933258,
+      "eval_runtime": 0.4727,
+      "eval_samples_per_second": 291.912,
+      "step": 300
+    },
     {
       "epoch": 16.0,
       "eval_accuracy": 78.26086956521739,
       "eval_average_metrics": 78.26086956521739,
       "eval_loss": 0.24881036579608917,
-      "eval_runtime": 0.3946,
-      "eval_samples_per_second": 349.683,
+      "eval_runtime": 0.474,
+      "eval_samples_per_second": 291.12,
       "step": 400
     },
     {
@@ -31,14 +49,23 @@
       "loss": 0.7619,
       "step": 500
     },
+    {
+      "epoch": 20.0,
+      "eval_accuracy": 76.08695652173914,
+      "eval_average_metrics": 76.08695652173914,
+      "eval_loss": 0.3080892562866211,
+      "eval_runtime": 0.5158,
+      "eval_samples_per_second": 267.545,
+      "step": 500
+    },
     {
       "epoch": 20.0,
       "step": 500,
       "total_flos": 7650655646515200.0,
       "train_loss": 0.7619044189453125,
-      "train_runtime": 190.0328,
-      "train_samples_per_second": 262.06,
-      "train_steps_per_second": 2.631
+      "train_runtime": 235.5565,
+      "train_samples_per_second": 211.414,
+      "train_steps_per_second": 2.123
     }
   ],
   "max_steps": 500,

training_args.bin

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:e5cd1b6dc14d58aac88259d963a6c2496413cfecbb9ff19605486a66162852e9
+oid sha256:2df0c1c1741f8ab42c8bcb5b0a2436f5275ba72a6f32b8533f2f0d10c2728a9f
 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": "rte", "eval_steps": 100, "evaluation_strategy": "steps", "greater_is_better": true, "learning_rate": 0.0003, "load_best_model_at_end": true, "max_source_length": 128, "metric_for_best_model": "average_metrics", "model_name_or_path": "../../../../plm_cache/t5-base", "num_train_epochs": 20, "output_dir": "outputs/bitfit/t5-base/rte", "overwrite_output_dir": true, "per_device_eval_batch_size": 100, "per_device_train_batch_size": 100, "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": "rte", "test_dataset_config_name": ["en"], "test_dataset_name": "rte", "tokenizer_name": "../../../../plm_cache/t5-base", "unfrozen_modules": ["deltas", "layer_norm", "final_layer_norm"], "warmup_steps": 500}
+{"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": "rte", "eval_steps": 100, "evaluation_strategy": "steps", "greater_is_better": true, "learning_rate": 0.0003, "load_best_model_at_end": true, "max_source_length": 128, "metric_for_best_model": "average_metrics", "model_name_or_path": "../../../../plm_cache/t5-base", "num_train_epochs": 20, "output_dir": "outputs/bitfit/t5-base/rte", "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": "rte", "test_dataset_config_name": ["en"], "test_dataset_name": "rte", "tokenizer_name": "../../../../plm_cache/t5-base", "unfrozen_modules": ["deltas", "layer_norm", "final_layer_norm"], "warmup_steps": 0}