fix model

all_results.json

@@ -1,6 +1,14 @@
 {
     "epoch": 3.0,
-    "train_runtime": 4486.3535,
+    "init_mem_cpu_alloc_delta": 2042978304,
+    "init_mem_cpu_peaked_delta": 121016320,
+    "init_mem_gpu_alloc_delta": 165904896,
+    "init_mem_gpu_peaked_delta": 0,
+    "train_mem_cpu_alloc_delta": 1035243520,
+    "train_mem_cpu_peaked_delta": 125026304,
+    "train_mem_gpu_alloc_delta": 497286656,
+    "train_mem_gpu_peaked_delta": 9670125056,
+    "train_runtime": 4763.7655,
     "train_samples": 200000,
-    "train_samples_per_second": 4.179
+    "train_samples_per_second": 3.936
 }

config.json

@@ -1,5 +1,5 @@
 {
-  "_name_or_path": "checkpoints/pretraining_checkpoints/summarization_dataset/medium_bert/",
+  "_name_or_path": "checkpoints/pretraining_checkpoints/translation_dataset/medium_bert/",
   "architectures": [
     "BertForSequenceClassification"
   ],

pytorch_model.bin

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:568b43af52bae16bd888b5287358b1e3c9a9cc651b989e415df3f5b031d84928
+oid sha256:c9ccf06ed86038a8d9be5679261a9cfe5282906b64626ffaaa2adb981d2938d3
 size 165557001

tokenizer_config.json

@@ -1 +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, "special_tokens_map_file": null, "name_or_path": "checkpoints/pretraining_checkpoints/summarization_dataset/medium_bert/", "do_basic_tokenize": true, "never_split": null}
+{"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, "special_tokens_map_file": null, "name_or_path": "checkpoints/pretraining_checkpoints/translation_dataset/medium_bert/", "do_basic_tokenize": true, "never_split": null}

train_results.json

@@ -1,6 +1,14 @@
 {
     "epoch": 3.0,
-    "train_runtime": 4486.3535,
+    "init_mem_cpu_alloc_delta": 2042978304,
+    "init_mem_cpu_peaked_delta": 121016320,
+    "init_mem_gpu_alloc_delta": 165904896,
+    "init_mem_gpu_peaked_delta": 0,
+    "train_mem_cpu_alloc_delta": 1035243520,
+    "train_mem_cpu_peaked_delta": 125026304,
+    "train_mem_gpu_alloc_delta": 497286656,
+    "train_mem_gpu_peaked_delta": 9670125056,
+    "train_runtime": 4763.7655,
     "train_samples": 200000,
-    "train_samples_per_second": 4.179
+    "train_samples_per_second": 3.936
 }

trainer_state.json

@@ -9,232 +9,232 @@
   "log_history": [
     {
       "epoch": 0.08,
-      "learning_rate": 8.888888888888888e-06,
-      "loss": 0.0494,
+      "learning_rate": 2.2222222222222223e-05,
+      "loss": 0.0104,
       "step": 500
     },
     {
       "epoch": 0.16,
-      "learning_rate": 1.7777777777777777e-05,
-      "loss": 0.0015,
+      "learning_rate": 4.4444444444444447e-05,
+      "loss": 0.0027,
       "step": 1000
     },
     {
       "epoch": 0.24,
-      "learning_rate": 1.9574468085106384e-05,
-      "loss": 0.0011,
+      "learning_rate": 4.893617021276596e-05,
+      "loss": 0.0026,
       "step": 1500
     },
     {
       "epoch": 0.32,
-      "learning_rate": 1.9007092198581563e-05,
-      "loss": 0.001,
+      "learning_rate": 4.751773049645391e-05,
+      "loss": 0.0024,
       "step": 2000
     },
     {
       "epoch": 0.4,
-      "learning_rate": 1.843971631205674e-05,
-      "loss": 0.0009,
+      "learning_rate": 4.609929078014185e-05,
+      "loss": 0.0026,
       "step": 2500
     },
     {
       "epoch": 0.48,
-      "learning_rate": 1.7872340425531915e-05,
-      "loss": 0.0008,
+      "learning_rate": 4.468085106382979e-05,
+      "loss": 0.0025,
       "step": 3000
     },
     {
       "epoch": 0.56,
-      "learning_rate": 1.7304964539007094e-05,
-      "loss": 0.0008,
+      "learning_rate": 4.3262411347517734e-05,
+      "loss": 0.0025,
       "step": 3500
     },
     {
       "epoch": 0.64,
-      "learning_rate": 1.673758865248227e-05,
-      "loss": 0.0008,
+      "learning_rate": 4.1843971631205674e-05,
+      "loss": 0.0025,
       "step": 4000
     },
     {
       "epoch": 0.72,
-      "learning_rate": 1.6170212765957446e-05,
-      "loss": 0.0008,
+      "learning_rate": 4.0425531914893614e-05,
+      "loss": 0.0024,
       "step": 4500
     },
     {
       "epoch": 0.8,
-      "learning_rate": 1.5602836879432626e-05,
-      "loss": 0.0007,
+      "learning_rate": 3.900709219858156e-05,
+      "loss": 0.0024,
       "step": 5000
     },
     {
       "epoch": 0.88,
-      "learning_rate": 1.5035460992907802e-05,
-      "loss": 0.0007,
+      "learning_rate": 3.75886524822695e-05,
+      "loss": 0.0023,
       "step": 5500
     },
     {
       "epoch": 0.96,
-      "learning_rate": 1.4468085106382981e-05,
-      "loss": 0.0007,
+      "learning_rate": 3.617021276595745e-05,
+      "loss": 0.0023,
       "step": 6000
     },
     {
       "epoch": 1.04,
-      "learning_rate": 1.3900709219858157e-05,
-      "loss": 0.0006,
+      "learning_rate": 3.4751773049645395e-05,
+      "loss": 0.002,
       "step": 6500
     },
     {
       "epoch": 1.12,
-      "learning_rate": 1.3333333333333333e-05,
-      "loss": 0.0005,
+      "learning_rate": 3.3333333333333335e-05,
+      "loss": 0.0015,
       "step": 7000
     },
     {
       "epoch": 1.2,
-      "learning_rate": 1.2765957446808513e-05,
-      "loss": 0.0006,
+      "learning_rate": 3.191489361702128e-05,
+      "loss": 0.0015,
       "step": 7500
     },
     {
       "epoch": 1.28,
-      "learning_rate": 1.2198581560283689e-05,
-      "loss": 0.0005,
+      "learning_rate": 3.0496453900709222e-05,
+      "loss": 0.0015,
       "step": 8000
     },
     {
       "epoch": 1.36,
-      "learning_rate": 1.1631205673758865e-05,
-      "loss": 0.0005,
+      "learning_rate": 2.9078014184397162e-05,
+      "loss": 0.0015,
       "step": 8500
     },
     {
       "epoch": 1.44,
-      "learning_rate": 1.1063829787234044e-05,
-      "loss": 0.0005,
+      "learning_rate": 2.765957446808511e-05,
+      "loss": 0.0015,
       "step": 9000
     },
     {
       "epoch": 1.52,
-      "learning_rate": 1.049645390070922e-05,
-      "loss": 0.0005,
+      "learning_rate": 2.6243971631205678e-05,
+      "loss": 0.0015,
       "step": 9500
     },
     {
       "epoch": 1.6,
-      "learning_rate": 9.929078014184398e-06,
-      "loss": 0.0005,
+      "learning_rate": 2.4825531914893618e-05,
+      "loss": 0.0015,
       "step": 10000
     },
     {
       "epoch": 1.68,
-      "learning_rate": 9.362836879432625e-06,
-      "loss": 0.0005,
+      "learning_rate": 2.340709219858156e-05,
+      "loss": 0.0015,
       "step": 10500
     },
     {
       "epoch": 1.76,
-      "learning_rate": 8.795460992907802e-06,
-      "loss": 0.0005,
+      "learning_rate": 2.1988652482269505e-05,
+      "loss": 0.0015,
       "step": 11000
     },
     {
       "epoch": 1.84,
-      "learning_rate": 8.22808510638298e-06,
-      "loss": 0.0005,
+      "learning_rate": 2.057021276595745e-05,
+      "loss": 0.0014,
       "step": 11500
     },
     {
       "epoch": 1.92,
-      "learning_rate": 7.660709219858156e-06,
-      "loss": 0.0005,
+      "learning_rate": 1.9151773049645392e-05,
+      "loss": 0.0014,
       "step": 12000
     },
     {
       "epoch": 2.0,
-      "learning_rate": 7.094468085106384e-06,
-      "loss": 0.0005,
+      "learning_rate": 1.7733333333333335e-05,
+      "loss": 0.0014,
       "step": 12500
     },
     {
       "epoch": 2.08,
-      "learning_rate": 6.5270921985815606e-06,
-      "loss": 0.0004,
+      "learning_rate": 1.6314893617021275e-05,
+      "loss": 0.001,
       "step": 13000
     },
     {
       "epoch": 2.16,
-      "learning_rate": 5.959716312056738e-06,
-      "loss": 0.0004,
+      "learning_rate": 1.4899290780141845e-05,
+      "loss": 0.001,
       "step": 13500
     },
     {
       "epoch": 2.24,
-      "learning_rate": 5.392340425531916e-06,
-      "loss": 0.0004,
+      "learning_rate": 1.3480851063829788e-05,
+      "loss": 0.001,
       "step": 14000
     },
     {
       "epoch": 2.32,
-      "learning_rate": 4.826099290780143e-06,
-      "loss": 0.0004,
+      "learning_rate": 1.2062411347517732e-05,
+      "loss": 0.001,
       "step": 14500
     },
     {
       "epoch": 2.4,
-      "learning_rate": 4.25872340425532e-06,
-      "loss": 0.0004,
+      "learning_rate": 1.0643971631205673e-05,
+      "loss": 0.001,
       "step": 15000
     },
     {
       "epoch": 2.48,
-      "learning_rate": 3.691347517730497e-06,
-      "loss": 0.0004,
+      "learning_rate": 9.228368794326241e-06,
+      "loss": 0.001,
       "step": 15500
     },
     {
       "epoch": 2.56,
-      "learning_rate": 3.1239716312056738e-06,
-      "loss": 0.0004,
+      "learning_rate": 7.809929078014185e-06,
+      "loss": 0.001,
       "step": 16000
     },
     {
       "epoch": 2.64,
-      "learning_rate": 2.5565957446808515e-06,
-      "loss": 0.0004,
+      "learning_rate": 6.391489361702129e-06,
+      "loss": 0.0009,
       "step": 16500
     },
     {
       "epoch": 2.72,
-      "learning_rate": 1.9892198581560283e-06,
-      "loss": 0.0004,
+      "learning_rate": 4.973049645390071e-06,
+      "loss": 0.0009,
       "step": 17000
     },
     {
       "epoch": 2.8,
-      "learning_rate": 1.4218439716312058e-06,
-      "loss": 0.0004,
+      "learning_rate": 3.557446808510639e-06,
+      "loss": 0.0009,
       "step": 17500
     },
     {
       "epoch": 2.88,
-      "learning_rate": 8.54468085106383e-07,
-      "loss": 0.0004,
+      "learning_rate": 2.1390070921985814e-06,
+      "loss": 0.0009,
       "step": 18000
     },
     {
       "epoch": 2.96,
-      "learning_rate": 2.870921985815603e-07,
-      "loss": 0.0004,
+      "learning_rate": 7.205673758865249e-07,
+      "loss": 0.0009,
       "step": 18500
     },
     {
       "epoch": 3.0,
       "step": 18750,
       "total_flos": 0,
-      "train_runtime": 4486.3535,
-      "train_samples_per_second": 4.179
+      "train_runtime": 4763.7655,
+      "train_samples_per_second": 3.936
     }
   ],
   "max_steps": 18750,

training_args.bin

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:9a0e800d9edf3fdf5c9726cdf7c0dd495bf7f0e0b3d11887556b9bc89f2bbbf4
+oid sha256:1c661bd4e99ec3e576c861e36134129d1f4746165ae4d7cefa82ed887f86b5db
 size 2479