[Model] roberta-base-qnli

config.json

@@ -0,0 +1,37 @@
+{
+  "_name_or_path": "/mnt/lustre/zhangyunchen/transformers/roberta-base",
+  "architectures": [
+    "RobertaForSequenceClassification"
+  ],
+  "attention_probs_dropout_prob": 0.1,
+  "bos_token_id": 0,
+  "classifier_dropout": null,
+  "eos_token_id": 2,
+  "finetuning_task": "qnli",
+  "hidden_act": "gelu",
+  "hidden_dropout_prob": 0.1,
+  "hidden_size": 768,
+  "id2label": {
+    "0": "entailment",
+    "1": "not_entailment"
+  },
+  "initializer_range": 0.02,
+  "intermediate_size": 3072,
+  "label2id": {
+    "entailment": 0,
+    "not_entailment": 1
+  },
+  "layer_norm_eps": 1e-05,
+  "max_position_embeddings": 514,
+  "model_type": "roberta",
+  "num_attention_heads": 12,
+  "num_hidden_layers": 12,
+  "pad_token_id": 1,
+  "position_embedding_type": "absolute",
+  "problem_type": "single_label_classification",
+  "torch_dtype": "float32",
+  "transformers_version": "4.16.2",
+  "type_vocab_size": 1,
+  "use_cache": true,
+  "vocab_size": 50265
+}

optimizer.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:8ba94e1b9f8291157f2f4782ed87b3c088c1477104b0fc7f93faeba6f73df97c
+size 997304355

pytorch_model.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:a2690e3b2c07a985122d73821a19ba7d088f4fb8d3871d256b167251f1b2e1ef
+size 498679699

rng_state.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:2d9efd3f53435a5a9d3b83112bc00936f822605e04266dadce57314cc319cc4e
+size 14659

scheduler.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:923c9fca46a8abfed1ee9d7e6f9222173fe0b5cccb71b39692914b9709f13c80
+size 623

special_tokens_map.json

@@ -0,0 +1 @@
+{"bos_token": "<s>", "eos_token": "</s>", "unk_token": "<unk>", "sep_token": "</s>", "pad_token": "<pad>", "cls_token": "<s>", "mask_token": {"content": "<mask>", "single_word": false, "lstrip": true, "rstrip": false, "normalized": false}}

tokenizer_config.json

@@ -0,0 +1 @@
+{"unk_token": "<unk>", "bos_token": "<s>", "eos_token": "</s>", "add_prefix_space": false, "errors": "replace", "sep_token": "</s>", "cls_token": "<s>", "pad_token": "<pad>", "mask_token": "<mask>", "trim_offsets": true, "special_tokens_map_file": null, "name_or_path": "/mnt/lustre/zhangyunchen/transformers/roberta-base", "tokenizer_class": "RobertaTokenizer"}

trainer_state.json

@@ -0,0 +1,316 @@
+{
+  "best_metric": 0.9298920007321985,
+  "best_model_checkpoint": "./fp32/models/qnli-roberta-base/checkpoint-9000",
+  "epoch": 6.108735491753207,
+  "global_step": 20000,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.31,
+      "learning_rate": 5.0890585241730285e-06,
+      "loss": 0.5591,
+      "step": 1000
+    },
+    {
+      "epoch": 0.31,
+      "eval_accuracy": 0.8694856306058942,
+      "eval_loss": 0.31558236479759216,
+      "eval_runtime": 21.239,
+      "eval_samples_per_second": 257.216,
+      "eval_steps_per_second": 8.051,
+      "step": 1000
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.988627132412675e-06,
+      "loss": 0.3512,
+      "step": 2000
+    },
+    {
+      "epoch": 0.61,
+      "eval_accuracy": 0.9007871133077063,
+      "eval_loss": 0.2616254389286041,
+      "eval_runtime": 15.8012,
+      "eval_samples_per_second": 345.733,
+      "eval_steps_per_second": 10.822,
+      "step": 2000
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 9.663688058489034e-06,
+      "loss": 0.3075,
+      "step": 3000
+    },
+    {
+      "epoch": 0.92,
+      "eval_accuracy": 0.9126853377265239,
+      "eval_loss": 0.23381730914115906,
+      "eval_runtime": 15.8636,
+      "eval_samples_per_second": 344.372,
+      "eval_steps_per_second": 10.779,
+      "step": 3000
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 9.338748984565394e-06,
+      "loss": 0.2678,
+      "step": 4000
+    },
+    {
+      "epoch": 1.22,
+      "eval_accuracy": 0.920190371590701,
+      "eval_loss": 0.22113902866840363,
+      "eval_runtime": 18.4732,
+      "eval_samples_per_second": 295.725,
+      "eval_steps_per_second": 9.257,
+      "step": 4000
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 9.013809910641755e-06,
+      "loss": 0.2453,
+      "step": 5000
+    },
+    {
+      "epoch": 1.53,
+      "eval_accuracy": 0.9103056928427604,
+      "eval_loss": 0.221624493598938,
+      "eval_runtime": 15.1776,
+      "eval_samples_per_second": 359.939,
+      "eval_steps_per_second": 11.267,
+      "step": 5000
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 8.688870836718116e-06,
+      "loss": 0.2364,
+      "step": 6000
+    },
+    {
+      "epoch": 1.83,
+      "eval_accuracy": 0.923302214900238,
+      "eval_loss": 0.20321357250213623,
+      "eval_runtime": 14.6843,
+      "eval_samples_per_second": 372.03,
+      "eval_steps_per_second": 11.645,
+      "step": 6000
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 8.363931762794477e-06,
+      "loss": 0.2109,
+      "step": 7000
+    },
+    {
+      "epoch": 2.14,
+      "eval_accuracy": 0.9240344133260113,
+      "eval_loss": 0.20744949579238892,
+      "eval_runtime": 15.2273,
+      "eval_samples_per_second": 358.765,
+      "eval_steps_per_second": 11.23,
+      "step": 7000
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 8.038992688870837e-06,
+      "loss": 0.1861,
+      "step": 8000
+    },
+    {
+      "epoch": 2.44,
+      "eval_accuracy": 0.9253157605711148,
+      "eval_loss": 0.22822654247283936,
+      "eval_runtime": 14.5847,
+      "eval_samples_per_second": 374.572,
+      "eval_steps_per_second": 11.725,
+      "step": 8000
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 7.714053614947198e-06,
+      "loss": 0.1857,
+      "step": 9000
+    },
+    {
+      "epoch": 2.75,
+      "eval_accuracy": 0.9298920007321985,
+      "eval_loss": 0.20272189378738403,
+      "eval_runtime": 14.8122,
+      "eval_samples_per_second": 368.818,
+      "eval_steps_per_second": 11.545,
+      "step": 9000
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 7.389114541023559e-06,
+      "loss": 0.1741,
+      "step": 10000
+    },
+    {
+      "epoch": 3.05,
+      "eval_accuracy": 0.9174446274940509,
+      "eval_loss": 0.2495354562997818,
+      "eval_runtime": 14.8558,
+      "eval_samples_per_second": 367.736,
+      "eval_steps_per_second": 11.511,
+      "step": 10000
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 7.0641754670999195e-06,
+      "loss": 0.1442,
+      "step": 11000
+    },
+    {
+      "epoch": 3.36,
+      "eval_accuracy": 0.9242174629324547,
+      "eval_loss": 0.2571660578250885,
+      "eval_runtime": 18.4719,
+      "eval_samples_per_second": 295.746,
+      "eval_steps_per_second": 9.257,
+      "step": 11000
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 6.73923639317628e-06,
+      "loss": 0.1453,
+      "step": 12000
+    },
+    {
+      "epoch": 3.67,
+      "eval_accuracy": 0.9256818597840014,
+      "eval_loss": 0.23131424188613892,
+      "eval_runtime": 14.7976,
+      "eval_samples_per_second": 369.181,
+      "eval_steps_per_second": 11.556,
+      "step": 12000
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 6.41429731925264e-06,
+      "loss": 0.1458,
+      "step": 13000
+    },
+    {
+      "epoch": 3.97,
+      "eval_accuracy": 0.923851363719568,
+      "eval_loss": 0.25046634674072266,
+      "eval_runtime": 14.7742,
+      "eval_samples_per_second": 369.767,
+      "eval_steps_per_second": 11.574,
+      "step": 13000
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 6.089358245329001e-06,
+      "loss": 0.1195,
+      "step": 14000
+    },
+    {
+      "epoch": 4.28,
+      "eval_accuracy": 0.923851363719568,
+      "eval_loss": 0.28182199597358704,
+      "eval_runtime": 14.6725,
+      "eval_samples_per_second": 372.328,
+      "eval_steps_per_second": 11.654,
+      "step": 14000
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 5.7644191714053624e-06,
+      "loss": 0.1209,
+      "step": 15000
+    },
+    {
+      "epoch": 4.58,
+      "eval_accuracy": 0.9223869668680212,
+      "eval_loss": 0.293856680393219,
+      "eval_runtime": 15.1476,
+      "eval_samples_per_second": 360.65,
+      "eval_steps_per_second": 11.289,
+      "step": 15000
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 5.439480097481723e-06,
+      "loss": 0.12,
+      "step": 16000
+    },
+    {
+      "epoch": 4.89,
+      "eval_accuracy": 0.9245835621453414,
+      "eval_loss": 0.28331321477890015,
+      "eval_runtime": 17.008,
+      "eval_samples_per_second": 321.202,
+      "eval_steps_per_second": 10.054,
+      "step": 16000
+    },
+    {
+      "epoch": 5.19,
+      "learning_rate": 5.114541023558083e-06,
+      "loss": 0.1086,
+      "step": 17000
+    },
+    {
+      "epoch": 5.19,
+      "eval_accuracy": 0.9249496613582281,
+      "eval_loss": 0.3361791968345642,
+      "eval_runtime": 16.1915,
+      "eval_samples_per_second": 337.399,
+      "eval_steps_per_second": 10.561,
+      "step": 17000
+    },
+    {
+      "epoch": 5.5,
+      "learning_rate": 4.789601949634444e-06,
+      "loss": 0.0991,
+      "step": 18000
+    },
+    {
+      "epoch": 5.5,
+      "eval_accuracy": 0.923302214900238,
+      "eval_loss": 0.3357701301574707,
+      "eval_runtime": 15.4017,
+      "eval_samples_per_second": 354.702,
+      "eval_steps_per_second": 11.103,
+      "step": 18000
+    },
+    {
+      "epoch": 5.8,
+      "learning_rate": 4.4646628757108045e-06,
+      "loss": 0.1033,
+      "step": 19000
+    },
+    {
+      "epoch": 5.8,
+      "eval_accuracy": 0.9264140582097748,
+      "eval_loss": 0.36776551604270935,
+      "eval_runtime": 18.05,
+      "eval_samples_per_second": 302.659,
+      "eval_steps_per_second": 9.474,
+      "step": 19000
+    },
+    {
+      "epoch": 6.11,
+      "learning_rate": 4.139723801787165e-06,
+      "loss": 0.0952,
+      "step": 20000
+    },
+    {
+      "epoch": 6.11,
+      "eval_accuracy": 0.9267801574226615,
+      "eval_loss": 0.36395037174224854,
+      "eval_runtime": 20.1267,
+      "eval_samples_per_second": 271.431,
+      "eval_steps_per_second": 8.496,
+      "step": 20000
+    }
+  ],
+  "max_steps": 32740,
+  "num_train_epochs": 10,
+  "total_flos": 4.2087902193024e+16,
+  "trial_name": null,
+  "trial_params": null
+}

training_args.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:5d7e533e6f8c4ac84cdae12c2f7145c2b1bc77abace1df22bc26571bc3b6082a
+size 3055