Training in progress, step 200

all_results.json

@@ -1,20 +1,20 @@
 {
     "epoch": 3.0,
-    "eval_average_metrics": 58.41817631314976,
-    "eval_em": 49.88986784140969,
-    "eval_f1": 66.94648478488982,
-    "eval_loss": 0.2216569185256958,
-    "eval_runtime": 9.9222,
-    "eval_samples_per_second": 244.301,
-    "test_average_metrics": 59.672400265556156,
-    "test_em": 51.55555555555556,
-    "test_f1": 67.78924497555676,
-    "test_loss": 0.20578201115131378,
-    "test_runtime": 10.0457,
-    "test_samples_per_second": 241.296,
-    "train_loss": 0.23175367241920627,
-    "train_runtime": 518.4287,
+    "eval_average_metrics": 62.27396685546466,
+    "eval_em": 52.6431718061674,
+    "eval_f1": 71.90476190476191,
+    "eval_loss": 0.22205480933189392,
+    "eval_runtime": 24.5121,
+    "eval_samples_per_second": 98.89,
+    "test_average_metrics": 63.46612466124662,
+    "test_em": 54.44444444444444,
+    "test_f1": 72.48780487804879,
+    "test_loss": 0.19974233210086823,
+    "test_runtime": 25.265,
+    "test_samples_per_second": 95.943,
+    "train_loss": 0.18877839930180654,
+    "train_runtime": 1478.3404,
     "train_samples": 27243,
-    "train_samples_per_second": 157.648,
-    "train_steps_per_second": 4.93
+    "train_samples_per_second": 55.284,
+    "train_steps_per_second": 1.729
 }

eval_results.json

@@ -1,9 +1,9 @@
 {
     "epoch": 3.0,
-    "eval_average_metrics": 58.41817631314976,
-    "eval_em": 49.88986784140969,
-    "eval_f1": 66.94648478488982,
-    "eval_loss": 0.2216569185256958,
-    "eval_runtime": 9.9222,
-    "eval_samples_per_second": 244.301
+    "eval_average_metrics": 62.27396685546466,
+    "eval_em": 52.6431718061674,
+    "eval_f1": 71.90476190476191,
+    "eval_loss": 0.22205480933189392,
+    "eval_runtime": 24.5121,
+    "eval_samples_per_second": 98.89
 }

pytorch_model.bin

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:243c21ed376bddb99303311130f2702572e78038275c8e8f2b67f5415c75b21d
-size 7551621
+oid sha256:f1ab444282eb590cc07dbabd31e47b50243e70ffccd042252568fb62a630f70c
+size 2631685

runs/Feb01_01-13-39_node1/events.out.tfevents.1643649325.node1

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:92dc55d11c380836a8aab5afa602143e627ad356537afe934f4f94e2e161ccbc
-size 8948
+oid sha256:32102cbbdcd9c4a795b0aa34ba5c29813b31321cf7915c06cf138ba7acdb7670
+size 9821

runs/Feb01_01-13-39_node1/events.out.tfevents.1643650828.node1

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:77e1993e71805ef73e081a4862b663ff96c1aa1575808c78da39cff8d9ac408d
+size 764

runs/Feb02_15-41-24_node1/1643787763.8152487/events.out.tfevents.1643787763.node1

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:ef1270b5efe2e01fbeb1d60927c997ac38e79a588538ae80052edd4781dd4e59
+size 5050

runs/Feb02_15-41-24_node1/events.out.tfevents.1643787763.node1

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:ddbeaefe73b2dfc4ea9ad1ad0465fab69d788e5c97f00a965dca9e048367f9af
+size 4338

test_results.json

@@ -1,9 +1,9 @@
 {
     "epoch": 3.0,
-    "test_average_metrics": 59.672400265556156,
-    "test_em": 51.55555555555556,
-    "test_f1": 67.78924497555676,
-    "test_loss": 0.20578201115131378,
-    "test_runtime": 10.0457,
-    "test_samples_per_second": 241.296
+    "test_average_metrics": 63.46612466124662,
+    "test_em": 54.44444444444444,
+    "test_f1": 72.48780487804879,
+    "test_loss": 0.19974233210086823,
+    "test_runtime": 25.265,
+    "test_samples_per_second": 95.943
 }

train_results.json

@@ -1,8 +1,8 @@
 {
     "epoch": 3.0,
-    "train_loss": 0.23175367241920627,
-    "train_runtime": 518.4287,
+    "train_loss": 0.18877839930180654,
+    "train_runtime": 1478.3404,
     "train_samples": 27243,
-    "train_samples_per_second": 157.648,
-    "train_steps_per_second": 4.93
+    "train_samples_per_second": 55.284,
+    "train_steps_per_second": 1.729
 }

trainer_state.json

@@ -1,105 +1,175 @@
 {
-  "best_metric": 58.41817631314976,
-  "best_model_checkpoint": "outputs/bitfit/t5-base/superglue-multirc/checkpoint-500",
+  "best_metric": 62.27396685546466,
+  "best_model_checkpoint": "outputs/bitfit/t5-base/superglue-multirc/checkpoint-2400",
   "epoch": 3.0,
   "global_step": 2556,
   "is_hyper_param_search": false,
   "is_local_process_zero": true,
   "is_world_process_zero": true,
   "log_history": [
+    {
+      "epoch": 0.23,
+      "eval_average_metrics": 58.45389071733288,
+      "eval_em": 49.88986784140969,
+      "eval_f1": 67.01791359325607,
+      "eval_loss": 0.21679049730300903,
+      "eval_runtime": 23.0565,
+      "eval_samples_per_second": 105.133,
+      "step": 200
+    },
+    {
+      "epoch": 0.47,
+      "eval_average_metrics": 61.27063054507724,
+      "eval_em": 51.98237885462555,
+      "eval_f1": 70.55888223552893,
+      "eval_loss": 0.2198609709739685,
+      "eval_runtime": 23.4946,
+      "eval_samples_per_second": 103.173,
+      "step": 400
+    },
     {
       "epoch": 0.59,
       "learning_rate": 0.0002413145539906103,
-      "loss": 0.3731,
+      "loss": 0.2582,
       "step": 500
     },
     {
-      "epoch": 0.59,
-      "eval_average_metrics": 58.41817631314976,
-      "eval_em": 49.88986784140969,
-      "eval_f1": 66.94648478488982,
-      "eval_loss": 0.2216569185256958,
-      "eval_runtime": 10.3072,
-      "eval_samples_per_second": 235.176,
-      "step": 500
+      "epoch": 0.7,
+      "eval_average_metrics": 59.16978670550705,
+      "eval_em": 49.779735682819386,
+      "eval_f1": 68.55983772819472,
+      "eval_loss": 0.20913197100162506,
+      "eval_runtime": 23.7161,
+      "eval_samples_per_second": 102.209,
+      "step": 600
+    },
+    {
+      "epoch": 0.94,
+      "eval_average_metrics": 61.270356888999785,
+      "eval_em": 51.43171806167401,
+      "eval_f1": 71.10899571632555,
+      "eval_loss": 0.20762182772159576,
+      "eval_runtime": 26.3745,
+      "eval_samples_per_second": 91.907,
+      "step": 800
     },
     {
       "epoch": 1.17,
       "learning_rate": 0.00018262910798122063,
-      "loss": 0.2051,
+      "loss": 0.1865,
       "step": 1000
     },
     {
       "epoch": 1.17,
-      "eval_average_metrics": 56.51724313400503,
-      "eval_em": 48.348017621145374,
-      "eval_f1": 64.68646864686468,
-      "eval_loss": 0.2394641786813736,
-      "eval_runtime": 10.1791,
-      "eval_samples_per_second": 238.134,
+      "eval_average_metrics": 61.45286279488806,
+      "eval_em": 51.87224669603524,
+      "eval_f1": 71.03347889374089,
+      "eval_loss": 0.2089201658964157,
+      "eval_runtime": 24.8833,
+      "eval_samples_per_second": 97.415,
       "step": 1000
     },
+    {
+      "epoch": 1.41,
+      "eval_average_metrics": 61.79826184816632,
+      "eval_em": 52.202643171806166,
+      "eval_f1": 71.39388052452648,
+      "eval_loss": 0.20319344103336334,
+      "eval_runtime": 25.4876,
+      "eval_samples_per_second": 95.105,
+      "step": 1200
+    },
+    {
+      "epoch": 1.64,
+      "eval_average_metrics": 62.13396621755072,
+      "eval_em": 52.53303964757709,
+      "eval_f1": 71.73489278752436,
+      "eval_loss": 0.21759574115276337,
+      "eval_runtime": 26.3526,
+      "eval_samples_per_second": 91.983,
+      "step": 1400
+    },
     {
       "epoch": 1.76,
       "learning_rate": 0.00012394366197183098,
-      "loss": 0.1986,
+      "loss": 0.1745,
       "step": 1500
     },
     {
-      "epoch": 1.76,
-      "eval_average_metrics": 57.765320466809,
-      "eval_em": 48.89867841409691,
-      "eval_f1": 66.6319625195211,
-      "eval_loss": 0.21828946471214294,
-      "eval_runtime": 10.2337,
-      "eval_samples_per_second": 236.864,
-      "step": 1500
+      "epoch": 1.88,
+      "eval_average_metrics": 62.159955413411296,
+      "eval_em": 52.09251101321586,
+      "eval_f1": 72.22739981360672,
+      "eval_loss": 0.21236789226531982,
+      "eval_runtime": 22.8791,
+      "eval_samples_per_second": 105.948,
+      "step": 1600
+    },
+    {
+      "epoch": 2.11,
+      "eval_average_metrics": 61.48864179183695,
+      "eval_em": 51.651982378854626,
+      "eval_f1": 71.32530120481927,
+      "eval_loss": 0.2192794233560562,
+      "eval_runtime": 23.8863,
+      "eval_samples_per_second": 101.481,
+      "step": 1800
     },
     {
       "epoch": 2.35,
       "learning_rate": 6.525821596244131e-05,
-      "loss": 0.1959,
+      "loss": 0.1692,
       "step": 2000
     },
     {
       "epoch": 2.35,
-      "eval_average_metrics": 57.370821730262364,
-      "eval_em": 48.56828193832599,
-      "eval_f1": 66.17336152219873,
-      "eval_loss": 0.2225608378648758,
-      "eval_runtime": 10.1351,
-      "eval_samples_per_second": 239.168,
+      "eval_average_metrics": 62.06655934464116,
+      "eval_em": 52.75330396475771,
+      "eval_f1": 71.37981472452462,
+      "eval_loss": 0.2168177217245102,
+      "eval_runtime": 23.5217,
+      "eval_samples_per_second": 103.054,
       "step": 2000
     },
     {
-      "epoch": 2.93,
-      "learning_rate": 6.5727699530516425e-06,
-      "loss": 0.1905,
-      "step": 2500
+      "epoch": 2.58,
+      "eval_average_metrics": 62.263646102553764,
+      "eval_em": 52.31277533039648,
+      "eval_f1": 72.21451687471105,
+      "eval_loss": 0.22297202050685883,
+      "eval_runtime": 22.3978,
+      "eval_samples_per_second": 108.225,
+      "step": 2200
+    },
+    {
+      "epoch": 2.82,
+      "eval_average_metrics": 62.27396685546466,
+      "eval_em": 52.6431718061674,
+      "eval_f1": 71.90476190476191,
+      "eval_loss": 0.22205480933189392,
+      "eval_runtime": 23.6577,
+      "eval_samples_per_second": 102.461,
+      "step": 2400
     },
     {
       "epoch": 2.93,
-      "eval_average_metrics": 58.251227638040206,
-      "eval_em": 49.118942731277535,
-      "eval_f1": 67.38351254480287,
-      "eval_loss": 0.21952344477176666,
-      "eval_runtime": 10.193,
-      "eval_samples_per_second": 237.809,
+      "learning_rate": 6.5727699530516425e-06,
+      "loss": 0.1589,
       "step": 2500
     },
     {
       "epoch": 3.0,
       "step": 2556,
-      "total_flos": 2.4915591799658496e+16,
-      "train_loss": 0.23175367241920627,
-      "train_runtime": 518.4287,
-      "train_samples_per_second": 157.648,
-      "train_steps_per_second": 4.93
+      "total_flos": 2.511166406964019e+16,
+      "train_loss": 0.18877839930180654,
+      "train_runtime": 1478.3404,
+      "train_samples_per_second": 55.284,
+      "train_steps_per_second": 1.729
     }
   ],
   "max_steps": 2556,
   "num_train_epochs": 3,
-  "total_flos": 2.4915591799658496e+16,
+  "total_flos": 2.511166406964019e+16,
   "trial_name": null,
   "trial_params": null
 }

training_args.bin

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:7b69a875f82add78a43a9830f1b279342b8db3bc578b10f33b14442fb873835e
+oid sha256:83e35f5470e9729f322b58c3e0e205808f864a2886e2f0aa7c9a151f826d1f0f
 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-multirc", "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": 3, "output_dir": "outputs/bitfit/t5-base/superglue-multirc", "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-multirc", "test_dataset_config_name": ["en"], "test_dataset_name": "superglue-multirc", "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-multirc", "eval_steps": 200, "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": 3, "output_dir": "outputs/bitfit/t5-base/superglue-multirc", "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-multirc", "test_dataset_config_name": ["en"], "test_dataset_name": "superglue-multirc", "tokenizer_name": "../../../../plm_cache/t5-base", "unfrozen_modules": ["deltas", "layer_norm", "final_layer_norm"], "warmup_steps": 0}