Training in progress, step 200

all_results.json

@@ -0,0 +1,18 @@
+{
+    "epoch": 3.0,
+    "eval_accuracy": 85.93368592351506,
+    "eval_average_metrics": 85.93368592351506,
+    "eval_loss": 0.17226365208625793,
+    "eval_runtime": 30.5086,
+    "eval_samples_per_second": 322.27,
+    "test_accuracy": 85.43046357615894,
+    "test_average_metrics": 85.43046357615894,
+    "test_loss": 0.17781256139278412,
+    "test_runtime": 30.4602,
+    "test_samples_per_second": 322.224,
+    "train_loss": 0.23374057220749037,
+    "train_runtime": 6523.9156,
+    "train_samples": 392702,
+    "train_samples_per_second": 180.583,
+    "train_steps_per_second": 1.806
+}

eval_results.json

@@ -0,0 +1,8 @@
+{
+    "epoch": 3.0,
+    "eval_accuracy": 85.93368592351506,
+    "eval_average_metrics": 85.93368592351506,
+    "eval_loss": 0.17226365208625793,
+    "eval_runtime": 30.5086,
+    "eval_samples_per_second": 322.27
+}

pytorch_model.bin

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:767e03b519dded4027505d5c63513e001ce03148e2109abad47d86f2e3b4f0c0
-size 1084131
+oid sha256:e9a8572ffe3e4be8efe2e7900d64ec74be93e038b262f6e3c27671e416504a72
+size 7551621

runs/Feb01_01-43-50_node1/1643651287.9640045/events.out.tfevents.1643651288.node1

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:b454d90781fb3b95be412b45b204718aabffd7601b56902ceb8544b912df1312
+size 5011

runs/Feb01_01-43-50_node1/events.out.tfevents.1643651287.node1

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:33aef57c063ff45fa04b0c31ecaa88eaaa4353be61bace4b357b68448b17722e
+size 4299

runs/Jan31_20-32-40_node1/events.out.tfevents.1643632563.node1

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:e0c9c0c26565d2d1658008ad6de47b2ab536f84f2212cf7aa53d7b767b679ec5
-size 14998
+oid sha256:c5aa4d4837e8f69fb3a26fae354bcc45114efcf19daef12bf0900a0d439be78b
+size 15674

runs/Jan31_20-32-40_node1/events.out.tfevents.1643639117.node1

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:59577aa34ff3b4403a54b05c504ef02355e1451d1ff1edb11544816937419304
+size 684

test_results.json

@@ -0,0 +1,8 @@
+{
+    "epoch": 3.0,
+    "test_accuracy": 85.43046357615894,
+    "test_average_metrics": 85.43046357615894,
+    "test_loss": 0.17781256139278412,
+    "test_runtime": 30.4602,
+    "test_samples_per_second": 322.224
+}

train_results.json

@@ -0,0 +1,8 @@
+{
+    "epoch": 3.0,
+    "train_loss": 0.23374057220749037,
+    "train_runtime": 6523.9156,
+    "train_samples": 392702,
+    "train_samples_per_second": 180.583,
+    "train_steps_per_second": 1.806
+}

trainer_state.json

@@ -0,0 +1,370 @@
+{
+  "best_metric": 85.93368592351506,
+  "best_model_checkpoint": "outputs/bitfit/t5-base/mnli/checkpoint-8500",
+  "epoch": 3.0,
+  "global_step": 11784,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0003,
+      "loss": 1.5813,
+      "step": 500
+    },
+    {
+      "epoch": 0.13,
+      "eval_accuracy": 78.88527257933279,
+      "eval_average_metrics": 78.88527257933279,
+      "eval_loss": 0.23019099235534668,
+      "eval_runtime": 42.8546,
+      "eval_samples_per_second": 229.427,
+      "step": 500
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0002867068415455512,
+      "loss": 0.2292,
+      "step": 1000
+    },
+    {
+      "epoch": 0.25,
+      "eval_accuracy": 84.4894222945484,
+      "eval_average_metrics": 84.4894222945484,
+      "eval_loss": 0.18283487856388092,
+      "eval_runtime": 37.9691,
+      "eval_samples_per_second": 258.947,
+      "step": 1000
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0002734136830911024,
+      "loss": 0.1956,
+      "step": 1500
+    },
+    {
+      "epoch": 0.38,
+      "eval_accuracy": 84.54027664768104,
+      "eval_average_metrics": 84.54027664768104,
+      "eval_loss": 0.1822226196527481,
+      "eval_runtime": 41.1857,
+      "eval_samples_per_second": 238.724,
+      "step": 1500
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00026012052463665364,
+      "loss": 0.1863,
+      "step": 2000
+    },
+    {
+      "epoch": 0.51,
+      "eval_accuracy": 85.22172497965825,
+      "eval_average_metrics": 85.22172497965825,
+      "eval_loss": 0.17375251650810242,
+      "eval_runtime": 40.7516,
+      "eval_samples_per_second": 241.267,
+      "step": 2000
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00024682736618220487,
+      "loss": 0.185,
+      "step": 2500
+    },
+    {
+      "epoch": 0.64,
+      "eval_accuracy": 85.0793327908869,
+      "eval_average_metrics": 85.0793327908869,
+      "eval_loss": 0.17669464647769928,
+      "eval_runtime": 30.4483,
+      "eval_samples_per_second": 322.908,
+      "step": 2500
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0002335342077277561,
+      "loss": 0.1775,
+      "step": 3000
+    },
+    {
+      "epoch": 0.76,
+      "eval_accuracy": 85.37428803905614,
+      "eval_average_metrics": 85.37428803905614,
+      "eval_loss": 0.17255689203739166,
+      "eval_runtime": 42.9629,
+      "eval_samples_per_second": 228.848,
+      "step": 3000
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.0002202410492733073,
+      "loss": 0.1769,
+      "step": 3500
+    },
+    {
+      "epoch": 0.89,
+      "eval_accuracy": 85.31326281529698,
+      "eval_average_metrics": 85.31326281529698,
+      "eval_loss": 0.17830435931682587,
+      "eval_runtime": 43.1763,
+      "eval_samples_per_second": 227.718,
+      "step": 3500
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.00020694789081885853,
+      "loss": 0.1785,
+      "step": 4000
+    },
+    {
+      "epoch": 1.02,
+      "eval_accuracy": 85.13018714401953,
+      "eval_average_metrics": 85.13018714401953,
+      "eval_loss": 0.1796959936618805,
+      "eval_runtime": 42.9567,
+      "eval_samples_per_second": 228.882,
+      "step": 4000
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.00019365473236440978,
+      "loss": 0.1732,
+      "step": 4500
+    },
+    {
+      "epoch": 1.15,
+      "eval_accuracy": 85.17087062652563,
+      "eval_average_metrics": 85.17087062652563,
+      "eval_loss": 0.17632290720939636,
+      "eval_runtime": 43.5824,
+      "eval_samples_per_second": 225.596,
+      "step": 4500
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.000180361573909961,
+      "loss": 0.171,
+      "step": 5000
+    },
+    {
+      "epoch": 1.27,
+      "eval_accuracy": 85.48616761594793,
+      "eval_average_metrics": 85.48616761594793,
+      "eval_loss": 0.17681637406349182,
+      "eval_runtime": 43.3608,
+      "eval_samples_per_second": 226.748,
+      "step": 5000
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.0001670684154555122,
+      "loss": 0.1721,
+      "step": 5500
+    },
+    {
+      "epoch": 1.4,
+      "eval_accuracy": 85.50650935720098,
+      "eval_average_metrics": 85.50650935720098,
+      "eval_loss": 0.17369242012500763,
+      "eval_runtime": 43.0099,
+      "eval_samples_per_second": 228.599,
+      "step": 5500
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.00015377525700106343,
+      "loss": 0.1691,
+      "step": 6000
+    },
+    {
+      "epoch": 1.53,
+      "eval_accuracy": 85.76078112286412,
+      "eval_average_metrics": 85.76078112286412,
+      "eval_loss": 0.1705426275730133,
+      "eval_runtime": 43.2716,
+      "eval_samples_per_second": 227.216,
+      "step": 6000
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.00014048209854661468,
+      "loss": 0.1692,
+      "step": 6500
+    },
+    {
+      "epoch": 1.65,
+      "eval_accuracy": 85.7099267697315,
+      "eval_average_metrics": 85.7099267697315,
+      "eval_loss": 0.1722770631313324,
+      "eval_runtime": 42.9316,
+      "eval_samples_per_second": 229.015,
+      "step": 6500
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.0001271889400921659,
+      "loss": 0.1673,
+      "step": 7000
+    },
+    {
+      "epoch": 1.78,
+      "eval_accuracy": 85.913344182262,
+      "eval_average_metrics": 85.913344182262,
+      "eval_loss": 0.17221254110336304,
+      "eval_runtime": 43.2906,
+      "eval_samples_per_second": 227.116,
+      "step": 7000
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.00011389578163771711,
+      "loss": 0.1674,
+      "step": 7500
+    },
+    {
+      "epoch": 1.91,
+      "eval_accuracy": 85.67941415785191,
+      "eval_average_metrics": 85.67941415785191,
+      "eval_loss": 0.1707369089126587,
+      "eval_runtime": 43.5484,
+      "eval_samples_per_second": 225.772,
+      "step": 7500
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 0.00010060262318326833,
+      "loss": 0.1666,
+      "step": 8000
+    },
+    {
+      "epoch": 2.04,
+      "eval_accuracy": 85.69975589910497,
+      "eval_average_metrics": 85.69975589910497,
+      "eval_loss": 0.1726333200931549,
+      "eval_runtime": 30.6332,
+      "eval_samples_per_second": 320.959,
+      "step": 8000
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 8.730946472881955e-05,
+      "loss": 0.1649,
+      "step": 8500
+    },
+    {
+      "epoch": 2.16,
+      "eval_accuracy": 85.93368592351506,
+      "eval_average_metrics": 85.93368592351506,
+      "eval_loss": 0.17226365208625793,
+      "eval_runtime": 30.5517,
+      "eval_samples_per_second": 321.816,
+      "step": 8500
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 7.401630627437078e-05,
+      "loss": 0.1652,
+      "step": 9000
+    },
+    {
+      "epoch": 2.29,
+      "eval_accuracy": 85.72009764035802,
+      "eval_average_metrics": 85.72009764035802,
+      "eval_loss": 0.17146219313144684,
+      "eval_runtime": 30.3221,
+      "eval_samples_per_second": 324.252,
+      "step": 9000
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 6.072314781992201e-05,
+      "loss": 0.1641,
+      "step": 9500
+    },
+    {
+      "epoch": 2.42,
+      "eval_accuracy": 85.66924328722538,
+      "eval_average_metrics": 85.66924328722538,
+      "eval_loss": 0.17002874612808228,
+      "eval_runtime": 30.4938,
+      "eval_samples_per_second": 322.427,
+      "step": 9500
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 4.742998936547323e-05,
+      "loss": 0.1626,
+      "step": 10000
+    },
+    {
+      "epoch": 2.55,
+      "eval_accuracy": 85.72009764035802,
+      "eval_average_metrics": 85.72009764035802,
+      "eval_loss": 0.17231160402297974,
+      "eval_runtime": 30.4513,
+      "eval_samples_per_second": 322.877,
+      "step": 10000
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 3.4136830911024454e-05,
+      "loss": 0.1648,
+      "step": 10500
+    },
+    {
+      "epoch": 2.67,
+      "eval_accuracy": 85.61838893409276,
+      "eval_average_metrics": 85.61838893409276,
+      "eval_loss": 0.17298805713653564,
+      "eval_runtime": 30.4131,
+      "eval_samples_per_second": 323.281,
+      "step": 10500
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.084367245657568e-05,
+      "loss": 0.1632,
+      "step": 11000
+    },
+    {
+      "epoch": 2.8,
+      "eval_accuracy": 85.62855980471929,
+      "eval_average_metrics": 85.62855980471929,
+      "eval_loss": 0.17419515550136566,
+      "eval_runtime": 30.8553,
+      "eval_samples_per_second": 318.648,
+      "step": 11000
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 7.550514002126905e-06,
+      "loss": 0.1632,
+      "step": 11500
+    },
+    {
+      "epoch": 2.93,
+      "eval_accuracy": 85.66924328722538,
+      "eval_average_metrics": 85.66924328722538,
+      "eval_loss": 0.17305442690849304,
+      "eval_runtime": 30.5078,
+      "eval_samples_per_second": 322.278,
+      "step": 11500
+    },
+    {
+      "epoch": 3.0,
+      "step": 11784,
+      "total_flos": 1.6728445568645094e+17,
+      "train_loss": 0.23374057220749037,
+      "train_runtime": 6523.9156,
+      "train_samples_per_second": 180.583,
+      "train_steps_per_second": 1.806
+    }
+  ],
+  "max_steps": 11784,
+  "num_train_epochs": 3,
+  "total_flos": 1.6728445568645094e+17,
+  "trial_name": null,
+  "trial_params": null
+}

training_args.bin

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:326faff2de82f88a3068b92c412261cd43afb80b76e57911821f923620f5e475
+oid sha256:3e63ea87be75b95629a7a1ac96920fa8d9c190a4c1cd01d194aa511d6abaff27
 size 3183

training_config.json

@@ -1 +1 @@
-{"dataset_config_name": ["en"], "delta_type": "bitfit", "do_eval": true, "do_test": true, "do_train": true, "eval_dataset_config_name": ["en"], "eval_dataset_name": "mnli", "eval_steps": 500, "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": 3, "output_dir": "outputs/bitfit/t5-base/mnli", "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": 500, "save_strategy": "steps", "save_total_limit": 1, "seed": 42, "split_validation_test": true, "task_name": "mnli", "test_dataset_config_name": ["en"], "test_dataset_name": "mnli", "tokenizer_name": "../../../../plm_cache/t5-base", "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": "mnli", "eval_steps": 200, "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": 3, "output_dir": "outputs/bitfit/t5-base/mnli", "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": "mnli", "test_dataset_config_name": ["en"], "test_dataset_name": "mnli", "tokenizer_name": "../../../../plm_cache/t5-base", "unfrozen_modules": ["deltas", "layer_norm", "final_layer_norm"], "warmup_steps": 0}