Training in progress, epoch 1

all_results.json

@@ -0,0 +1,18 @@
+{
+    "epoch": 20.0,
+    "eval_accuracy": 80.61162079510703,
+    "eval_average_metrics": 80.61162079510703,
+    "eval_loss": 0.3976394534111023,
+    "eval_runtime": 4.7604,
+    "eval_samples_per_second": 343.462,
+    "test_accuracy": 79.38837920489297,
+    "test_average_metrics": 79.38837920489297,
+    "test_loss": 0.4527607262134552,
+    "test_runtime": 4.773,
+    "test_samples_per_second": 342.554,
+    "train_loss": 0.1054745042930215,
+    "train_runtime": 1312.1786,
+    "train_samples": 9427,
+    "train_samples_per_second": 143.685,
+    "train_steps_per_second": 4.496
+}

eval_results.json

@@ -0,0 +1,8 @@
+{
+    "epoch": 20.0,
+    "eval_accuracy": 80.61162079510703,
+    "eval_average_metrics": 80.61162079510703,
+    "eval_loss": 0.3976394534111023,
+    "eval_runtime": 4.7604,
+    "eval_samples_per_second": 343.462
+}

pytorch_model.bin

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:e8ec286b6690ddc35938a4f54e8272579a13555f3e098bd237104857b4450147
-size 9913989
+oid sha256:8b456c85c073efe312e20263b9d0ea6131e6bd25a2ae380d0b5b9a05bb7e4db5
+size 7551621

runs/Jan30_11-25-30_node4/events.out.tfevents.1643513230.node4

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:bb09163ddf044a8f41816cfed3a1f519f932ee2d9f38c946fb9179dd56034412
-size 11358
+oid sha256:6b3097ca297a56184e6eb755a79457dd6e75db14cef86d5e8fc37e471d53a4d8
+size 12835

runs/Jan30_11-25-30_node4/events.out.tfevents.1643514547.node4

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:a06f71e8d0fe41bbd53112a1820f61108a923b473c799142afa506453972e8e0
+size 684

runs/Jan30_12-18-00_node4/1643516350.045366/events.out.tfevents.1643516350.node4

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:12e17722fdb52961e691e9ee8928edb3ec5574970545cf383485d153f3f472ce
+size 5020

runs/Jan30_12-18-00_node4/events.out.tfevents.1643516350.node4

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:4a0e0ab3d592eabf8ab8c1d2b33fd34919a882a5c791292a58adbac90aafed60
+size 4314

test_results.json

@@ -0,0 +1,8 @@
+{
+    "epoch": 20.0,
+    "test_accuracy": 79.38837920489297,
+    "test_average_metrics": 79.38837920489297,
+    "test_loss": 0.4527607262134552,
+    "test_runtime": 4.773,
+    "test_samples_per_second": 342.554
+}

train_results.json

@@ -0,0 +1,8 @@
+{
+    "epoch": 20.0,
+    "train_loss": 0.1054745042930215,
+    "train_runtime": 1312.1786,
+    "train_samples": 9427,
+    "train_samples_per_second": 143.685,
+    "train_steps_per_second": 4.496
+}

trainer_state.json

@@ -0,0 +1,271 @@
+{
+  "best_metric": 80.61162079510703,
+  "best_model_checkpoint": "outputs/adapter/superglue-boolq/checkpoint-3835",
+  "epoch": 20.0,
+  "global_step": 5900,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 1.0,
+      "eval_accuracy": 62.62996941896024,
+      "eval_average_metrics": 62.62996941896024,
+      "eval_loss": 0.2789694368839264,
+      "eval_runtime": 4.4788,
+      "eval_samples_per_second": 365.053,
+      "step": 295
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.00027457627118644066,
+      "loss": 0.3259,
+      "step": 500
+    },
+    {
+      "epoch": 2.0,
+      "eval_accuracy": 68.31804281345566,
+      "eval_average_metrics": 68.31804281345566,
+      "eval_loss": 0.2598051428794861,
+      "eval_runtime": 4.7399,
+      "eval_samples_per_second": 344.945,
+      "step": 590
+    },
+    {
+      "epoch": 3.0,
+      "eval_accuracy": 74.6788990825688,
+      "eval_average_metrics": 74.6788990825688,
+      "eval_loss": 0.22243507206439972,
+      "eval_runtime": 4.7583,
+      "eval_samples_per_second": 343.612,
+      "step": 885
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 0.00024915254237288135,
+      "loss": 0.2455,
+      "step": 1000
+    },
+    {
+      "epoch": 4.0,
+      "eval_accuracy": 74.6788990825688,
+      "eval_average_metrics": 74.6788990825688,
+      "eval_loss": 0.21567188203334808,
+      "eval_runtime": 4.7503,
+      "eval_samples_per_second": 344.185,
+      "step": 1180
+    },
+    {
+      "epoch": 5.0,
+      "eval_accuracy": 79.32721712538226,
+      "eval_average_metrics": 79.32721712538226,
+      "eval_loss": 0.2073381394147873,
+      "eval_runtime": 4.7337,
+      "eval_samples_per_second": 345.395,
+      "step": 1475
+    },
+    {
+      "epoch": 5.08,
+      "learning_rate": 0.000223728813559322,
+      "loss": 0.1872,
+      "step": 1500
+    },
+    {
+      "epoch": 6.0,
+      "eval_accuracy": 78.71559633027523,
+      "eval_average_metrics": 78.71559633027523,
+      "eval_loss": 0.207603320479393,
+      "eval_runtime": 4.7412,
+      "eval_samples_per_second": 344.851,
+      "step": 1770
+    },
+    {
+      "epoch": 6.78,
+      "learning_rate": 0.0001983050847457627,
+      "loss": 0.134,
+      "step": 2000
+    },
+    {
+      "epoch": 7.0,
+      "eval_accuracy": 80.0,
+      "eval_average_metrics": 80.0,
+      "eval_loss": 0.22447817027568817,
+      "eval_runtime": 4.7578,
+      "eval_samples_per_second": 343.643,
+      "step": 2065
+    },
+    {
+      "epoch": 8.0,
+      "eval_accuracy": 78.83792048929664,
+      "eval_average_metrics": 78.83792048929664,
+      "eval_loss": 0.2642129361629486,
+      "eval_runtime": 4.7785,
+      "eval_samples_per_second": 342.156,
+      "step": 2360
+    },
+    {
+      "epoch": 8.47,
+      "learning_rate": 0.0001728813559322034,
+      "loss": 0.0978,
+      "step": 2500
+    },
+    {
+      "epoch": 9.0,
+      "eval_accuracy": 80.1223241590214,
+      "eval_average_metrics": 80.1223241590214,
+      "eval_loss": 0.26512306928634644,
+      "eval_runtime": 4.7549,
+      "eval_samples_per_second": 343.857,
+      "step": 2655
+    },
+    {
+      "epoch": 10.0,
+      "eval_accuracy": 79.69418960244649,
+      "eval_average_metrics": 79.69418960244649,
+      "eval_loss": 0.2741082012653351,
+      "eval_runtime": 4.749,
+      "eval_samples_per_second": 344.283,
+      "step": 2950
+    },
+    {
+      "epoch": 10.17,
+      "learning_rate": 0.00014745762711864405,
+      "loss": 0.0743,
+      "step": 3000
+    },
+    {
+      "epoch": 11.0,
+      "eval_accuracy": 80.42813455657493,
+      "eval_average_metrics": 80.42813455657493,
+      "eval_loss": 0.31817376613616943,
+      "eval_runtime": 4.7525,
+      "eval_samples_per_second": 344.028,
+      "step": 3245
+    },
+    {
+      "epoch": 11.86,
+      "learning_rate": 0.00012203389830508474,
+      "loss": 0.0541,
+      "step": 3500
+    },
+    {
+      "epoch": 12.0,
+      "eval_accuracy": 79.81651376146789,
+      "eval_average_metrics": 79.81651376146789,
+      "eval_loss": 0.33367282152175903,
+      "eval_runtime": 4.7584,
+      "eval_samples_per_second": 343.602,
+      "step": 3540
+    },
+    {
+      "epoch": 13.0,
+      "eval_accuracy": 80.61162079510703,
+      "eval_average_metrics": 80.61162079510703,
+      "eval_loss": 0.3976394534111023,
+      "eval_runtime": 4.7841,
+      "eval_samples_per_second": 341.758,
+      "step": 3835
+    },
+    {
+      "epoch": 13.56,
+      "learning_rate": 9.661016949152541e-05,
+      "loss": 0.0399,
+      "step": 4000
+    },
+    {
+      "epoch": 14.0,
+      "eval_accuracy": 79.8776758409786,
+      "eval_average_metrics": 79.8776758409786,
+      "eval_loss": 0.41730642318725586,
+      "eval_runtime": 4.7467,
+      "eval_samples_per_second": 344.451,
+      "step": 4130
+    },
+    {
+      "epoch": 15.0,
+      "eval_accuracy": 80.18348623853211,
+      "eval_average_metrics": 80.18348623853211,
+      "eval_loss": 0.4511031210422516,
+      "eval_runtime": 4.7544,
+      "eval_samples_per_second": 343.89,
+      "step": 4425
+    },
+    {
+      "epoch": 15.25,
+      "learning_rate": 7.11864406779661e-05,
+      "loss": 0.0303,
+      "step": 4500
+    },
+    {
+      "epoch": 16.0,
+      "eval_accuracy": 79.63302752293579,
+      "eval_average_metrics": 79.63302752293579,
+      "eval_loss": 0.4779604375362396,
+      "eval_runtime": 4.7435,
+      "eval_samples_per_second": 344.682,
+      "step": 4720
+    },
+    {
+      "epoch": 16.95,
+      "learning_rate": 4.576271186440678e-05,
+      "loss": 0.0234,
+      "step": 5000
+    },
+    {
+      "epoch": 17.0,
+      "eval_accuracy": 80.18348623853211,
+      "eval_average_metrics": 80.18348623853211,
+      "eval_loss": 0.5195048451423645,
+      "eval_runtime": 4.7472,
+      "eval_samples_per_second": 344.414,
+      "step": 5015
+    },
+    {
+      "epoch": 18.0,
+      "eval_accuracy": 79.69418960244649,
+      "eval_average_metrics": 79.69418960244649,
+      "eval_loss": 0.5308667421340942,
+      "eval_runtime": 4.7851,
+      "eval_samples_per_second": 341.683,
+      "step": 5310
+    },
+    {
+      "epoch": 18.64,
+      "learning_rate": 2.0338983050847455e-05,
+      "loss": 0.019,
+      "step": 5500
+    },
+    {
+      "epoch": 19.0,
+      "eval_accuracy": 80.06116207951071,
+      "eval_average_metrics": 80.06116207951071,
+      "eval_loss": 0.5453023910522461,
+      "eval_runtime": 4.7413,
+      "eval_samples_per_second": 344.844,
+      "step": 5605
+    },
+    {
+      "epoch": 20.0,
+      "eval_accuracy": 80.36697247706422,
+      "eval_average_metrics": 80.36697247706422,
+      "eval_loss": 0.5537198185920715,
+      "eval_runtime": 4.7547,
+      "eval_samples_per_second": 343.869,
+      "step": 5900
+    },
+    {
+      "epoch": 20.0,
+      "step": 5900,
+      "total_flos": 5.80271991662592e+16,
+      "train_loss": 0.1054745042930215,
+      "train_runtime": 1312.1786,
+      "train_samples_per_second": 143.685,
+      "train_steps_per_second": 4.496
+    }
+  ],
+  "max_steps": 5900,
+  "num_train_epochs": 20,
+  "total_flos": 5.80271991662592e+16,
+  "trial_name": null,
+  "trial_params": null
+}

training_args.bin

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:ef2956b205d4416bd31ce0b90cc6a8f19c360cb74ad7084c14c7b6ce0fa8535c
+oid sha256:035e52789313ccd7d5fc94e73b66a6867aa661d40c1539a49dae23bb7e66974e
 size 3183

training_config.json

@@ -1 +1 @@
-{"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-boolq", "evaluation_strategy": "epoch", "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", "non_linearity": "gelu_new", "num_train_epochs": 20, "output_dir": "outputs/adapter/superglue-boolq", "per_device_eval_batch_size": 100, "per_device_train_batch_size": 32, "predict_with_generate": true, "save_strategy": "epoch", "save_total_limit": 1, "split_validation_test": true, "task_name": "superglue-boolq", "test_dataset_config_name": ["en"], "unfrozen_modules": ["deltas", "layer_norm", "final_layer_norm"], "test_dataset_name": "superglue-boolq", "tokenizer_name": "../../../../plm_cache/t5-base", "warmup_steps": 0, "overwrite_output_dir": true, "push_to_hub": true}
+{"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-boolq", "evaluation_strategy": "epoch", "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", "non_linearity": "gelu_new", "num_train_epochs": 20, "output_dir": "outputs/adapter/superglue-boolq", "per_device_eval_batch_size": 100, "per_device_train_batch_size": 32, "predict_with_generate": true, "save_strategy": "epoch", "save_total_limit": 1, "split_validation_test": true, "task_name": "superglue-boolq", "test_dataset_config_name": ["en"], "unfrozen_modules": ["deltas", "layer_norm", "final_layer_norm"], "bottleneck_dim": 24, "test_dataset_name": "superglue-boolq", "tokenizer_name": "../../../../plm_cache/t5-base", "warmup_steps": 0, "overwrite_output_dir": true, "push_to_hub": true}