update qnli

all_results.json

@@ -0,0 +1,23 @@
+{
+    "epoch": 3.0,
+    "eval_accuracy": 0.9295259015193117,
+    "eval_loss": 0.2601969242095947,
+    "eval_mem_cpu_alloc_delta": 8011776,
+    "eval_mem_cpu_peaked_delta": 122880,
+    "eval_mem_gpu_alloc_delta": 0,
+    "eval_mem_gpu_peaked_delta": 34744320,
+    "eval_runtime": 38.4335,
+    "eval_samples": 5463,
+    "eval_samples_per_second": 142.142,
+    "init_mem_cpu_alloc_delta": 3165298688,
+    "init_mem_cpu_peaked_delta": 319688704,
+    "init_mem_gpu_alloc_delta": 439072256,
+    "init_mem_gpu_peaked_delta": 0,
+    "train_mem_cpu_alloc_delta": 159789056,
+    "train_mem_cpu_peaked_delta": 197320704,
+    "train_mem_gpu_alloc_delta": 1379092992,
+    "train_mem_gpu_peaked_delta": 3560779776,
+    "train_runtime": 6184.2743,
+    "train_samples": 104743,
+    "train_samples_per_second": 1.588
+}

config.json

@@ -0,0 +1,28 @@
+{
+  "_name_or_path": "google/electra-base-discriminator",
+  "architectures": [
+    "ElectraForSequenceClassification"
+  ],
+  "attention_probs_dropout_prob": 0.1,
+  "embedding_size": 768,
+  "finetuning_task": "qnli",
+  "hidden_act": "gelu",
+  "hidden_dropout_prob": 0.1,
+  "hidden_size": 768,
+  "initializer_range": 0.02,
+  "intermediate_size": 3072,
+  "layer_norm_eps": 1e-12,
+  "max_position_embeddings": 512,
+  "model_type": "electra",
+  "num_attention_heads": 12,
+  "num_hidden_layers": 12,
+  "pad_token_id": 0,
+  "position_embedding_type": "absolute",
+  "summary_activation": "gelu",
+  "summary_last_dropout": 0.1,
+  "summary_type": "first",
+  "summary_use_proj": true,
+  "transformers_version": "4.6.0.dev0",
+  "type_vocab_size": 2,
+  "vocab_size": 30522
+}

eval_results.json

@@ -0,0 +1,12 @@
+{
+    "epoch": 3.0,
+    "eval_accuracy": 0.9295259015193117,
+    "eval_loss": 0.2601969242095947,
+    "eval_mem_cpu_alloc_delta": 8011776,
+    "eval_mem_cpu_peaked_delta": 122880,
+    "eval_mem_gpu_alloc_delta": 0,
+    "eval_mem_gpu_peaked_delta": 34744320,
+    "eval_runtime": 38.4335,
+    "eval_samples": 5463,
+    "eval_samples_per_second": 142.142
+}

pytorch_model.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:69a5006916636c422bd5496b9297b1cb6cbfb847b8db811fe2a5cf06229e364b
+size 438023702

special_tokens_map.json

@@ -0,0 +1 @@
+{"unk_token": "[UNK]", "sep_token": "[SEP]", "pad_token": "[PAD]", "cls_token": "[CLS]", "mask_token": "[MASK]"}

tokenizer_config.json

@@ -0,0 +1 @@
+{"do_lower_case": true, "unk_token": "[UNK]", "sep_token": "[SEP]", "pad_token": "[PAD]", "cls_token": "[CLS]", "mask_token": "[MASK]", "tokenize_chinese_chars": true, "strip_accents": null, "model_max_length": 512, "special_tokens_map_file": null, "name_or_path": "google/electra-base-discriminator"}

train_results.json

@@ -0,0 +1,14 @@
+{
+    "epoch": 3.0,
+    "init_mem_cpu_alloc_delta": 3165298688,
+    "init_mem_cpu_peaked_delta": 319688704,
+    "init_mem_gpu_alloc_delta": 439072256,
+    "init_mem_gpu_peaked_delta": 0,
+    "train_mem_cpu_alloc_delta": 159789056,
+    "train_mem_cpu_peaked_delta": 197320704,
+    "train_mem_gpu_alloc_delta": 1379092992,
+    "train_mem_gpu_peaked_delta": 3560779776,
+    "train_runtime": 6184.2743,
+    "train_samples": 104743,
+    "train_samples_per_second": 1.588
+}

trainer_state.json

@@ -0,0 +1,137 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 3.0,
+  "global_step": 9822,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8981877418041133e-05,
+      "loss": 0.4141,
+      "step": 500
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.7963754836082265e-05,
+      "loss": 0.311,
+      "step": 1000
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.69456322541234e-05,
+      "loss": 0.2751,
+      "step": 1500
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.592750967216453e-05,
+      "loss": 0.2735,
+      "step": 2000
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4909387090205662e-05,
+      "loss": 0.2614,
+      "step": 2500
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.3891264508246794e-05,
+      "loss": 0.2478,
+      "step": 3000
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.2873141926287927e-05,
+      "loss": 0.2108,
+      "step": 3500
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.1855019344329057e-05,
+      "loss": 0.179,
+      "step": 4000
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.083689676237019e-05,
+      "loss": 0.1689,
+      "step": 4500
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 9.818774180411322e-06,
+      "loss": 0.1651,
+      "step": 5000
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.800651598452455e-06,
+      "loss": 0.1701,
+      "step": 5500
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 7.782529016493586e-06,
+      "loss": 0.1715,
+      "step": 6000
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 6.764406434534719e-06,
+      "loss": 0.1662,
+      "step": 6500
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 5.74628385257585e-06,
+      "loss": 0.1148,
+      "step": 7000
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 4.728161270616982e-06,
+      "loss": 0.1101,
+      "step": 7500
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 3.7100386886581147e-06,
+      "loss": 0.1156,
+      "step": 8000
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 2.6919161066992467e-06,
+      "loss": 0.1107,
+      "step": 8500
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.6737935247403788e-06,
+      "loss": 0.1054,
+      "step": 9000
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 6.55670942781511e-07,
+      "loss": 0.107,
+      "step": 9500
+    },
+    {
+      "epoch": 3.0,
+      "step": 9822,
+      "total_flos": 2.6421487163260416e+16,
+      "train_runtime": 6184.2743,
+      "train_samples_per_second": 1.588
+    }
+  ],
+  "max_steps": 9822,
+  "num_train_epochs": 3,
+  "total_flos": 2.6421487163260416e+16,
+  "trial_name": null,
+  "trial_params": null
+}

training_args.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:b62a682d81e961812b3cae9d3a79c896ad58655cf5d92dd17acc19be4ec284a8
+size 2415