update training model

Files changed (15) hide show

README.md +14 -18
all_results.json +11 -11
config.json +8 -1
eval_results.json +7 -7
generation_config.json +225 -0
merges.txt +1 -1
pred.txt +1 -1
preprocessor_config.json +0 -0
pytorch_model.bin +1 -1
tokenizer.json +0 -0
tokenizer_config.json +0 -1
train_results.json +5 -5
trainer_state.json +21 -1255
training_args.bin +2 -2
vocab.json +0 -0

README.md CHANGED Viewed

@@ -1,6 +1,6 @@
 ---
 language:
-- zh
 license: apache-2.0
 tags:
 - whisper-event
@@ -10,7 +10,7 @@ datasets:
 metrics:
 - wer
 model-index:
-- name: Whisper large-v2 nan-tw
   results:
   - task:
       name: Automatic Speech Recognition
@@ -19,24 +19,24 @@ model-index:
       name: mozilla-foundation/common_voice_11_0 nan-tw
       type: mozilla-foundation/common_voice_11_0
       config: nan-tw
-      split: train
       args: nan-tw
     metrics:
     - name: Wer
       type: wer
-      value: 118.50381679389312
 ---
 <!-- This model card has been generated automatically according to the information the Trainer had access to. You
 should probably proofread and complete it, then remove this comment. -->
-# Whisper large-v2 nan-tw
 This model is a fine-tuned version of [openai/whisper-large-v2](https://huggingface.co/openai/whisper-large-v2) on the mozilla-foundation/common_voice_11_0 nan-tw dataset.
 It achieves the following results on the evaluation set:
-- Loss: 3.2129
-- Wer: 118.5038
-- Cer: 123.4531
 ## Model description
@@ -61,24 +61,20 @@ The following hyperparameters were used during training:
 - seed: 42
 - optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
 - lr_scheduler_type: linear
-- lr_scheduler_warmup_steps: 500
-- training_steps: 5000
 - mixed_precision_training: Native AMP
 ### Training results
-| Training Loss | Epoch | Step | Validation Loss | Wer      | Cer      |
-|:-------------:|:-----:|:----:|:---------------:|:--------:|:--------:|
-| 3.5696        | 1.04  | 1000 | 3.4190          | 96.8550  | 96.8910  |
-| 3.1453        | 2.08  | 2000 | 3.2383          | 98.9313  | 98.9436  |
-| 3.0722        | 3.13  | 3000 | 3.2043          | 129.0687 | 158.5270 |
-| 2.8327        | 5.01  | 4000 | 3.2258          | 327.9084 | 333.0516 |
-| 2.6468        | 6.05  | 5000 | 3.2129          | 118.5038 | 123.4531 |
 ### Framework versions
-- Transformers 4.25.1
 - Pytorch 1.13.1+cu117
 - Datasets 2.8.0
 - Tokenizers 0.13.2

 ---
 language:
+- nan-tw
 license: apache-2.0
 tags:
 - whisper-event
 metrics:
 - wer
 model-index:
+- name: Whisper large-v2 nan-tw only char
   results:
   - task:
       name: Automatic Speech Recognition
       name: mozilla-foundation/common_voice_11_0 nan-tw
       type: mozilla-foundation/common_voice_11_0
       config: nan-tw
+      split: test
       args: nan-tw
     metrics:
     - name: Wer
       type: wer
+      value: 48.06106870229008
 ---
 <!-- This model card has been generated automatically according to the information the Trainer had access to. You
 should probably proofread and complete it, then remove this comment. -->
+# vscode
 This model is a fine-tuned version of [openai/whisper-large-v2](https://huggingface.co/openai/whisper-large-v2) on the mozilla-foundation/common_voice_11_0 nan-tw dataset.
 It achieves the following results on the evaluation set:
+- Loss: 0.9147
+- Wer: 48.0611
+- Cer: 48.1739
 ## Model description
 - seed: 42
 - optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
 - lr_scheduler_type: linear
+- lr_scheduler_warmup_steps: 100
+- training_steps: 1000
 - mixed_precision_training: Native AMP
 ### Training results
+| Training Loss | Epoch | Step | Validation Loss | Wer     | Cer     |
+|:-------------:|:-----:|:----:|:---------------:|:-------:|:-------:|
+| 0.9198        | 1.38  | 1000 | 0.9147          | 48.0611 | 48.1739 |
 ### Framework versions
+- Transformers 4.27.0.dev0
 - Pytorch 1.13.1+cu117
 - Datasets 2.8.0
 - Tokenizers 0.13.2

all_results.json CHANGED Viewed

@@ -1,13 +1,13 @@
 {
-    "epoch": 6.05,
-    "eval_cer": 123.45306368849984,
-    "eval_loss": 3.212874174118042,
-    "eval_runtime": 303.2793,
-    "eval_samples_per_second": 3.251,
-    "eval_steps_per_second": 1.626,
-    "eval_wer": 118.50381679389312,
-    "train_loss": 3.0522214595794677,
-    "train_runtime": 5458.8622,
-    "train_samples_per_second": 1.832,
-    "train_steps_per_second": 0.916
 }

 {
+    "epoch": 1.38,
+    "eval_cer": 48.17386054935105,
+    "eval_loss": 0.9147109389305115,
+    "eval_runtime": 340.4629,
+    "eval_samples_per_second": 2.896,
+    "eval_steps_per_second": 1.448,
+    "eval_wer": 48.06106870229008,
+    "train_loss": 0.91982373046875,
+    "train_runtime": 1347.6358,
+    "train_samples_per_second": 1.484,
+    "train_steps_per_second": 0.742
 }

config.json CHANGED Viewed

@@ -2,6 +2,7 @@
   "_name_or_path": "openai/whisper-large-v2",
   "activation_dropout": 0.0,
   "activation_function": "gelu",
   "architectures": [
     "WhisperForConditionalGeneration"
   ],
@@ -26,6 +27,12 @@
   "forced_decoder_ids": null,
   "init_std": 0.02,
   "is_encoder_decoder": true,
   "max_length": 448,
   "max_source_positions": 1500,
   "max_target_positions": 448,
@@ -35,7 +42,7 @@
   "pad_token_id": 50257,
   "scale_embedding": false,
   "torch_dtype": "float32",
-  "transformers_version": "4.25.1",
   "use_cache": true,
   "vocab_size": 51865
 }

   "_name_or_path": "openai/whisper-large-v2",
   "activation_dropout": 0.0,
   "activation_function": "gelu",
+  "apply_spec_augment": false,
   "architectures": [
     "WhisperForConditionalGeneration"
   ],
   "forced_decoder_ids": null,
   "init_std": 0.02,
   "is_encoder_decoder": true,
+  "mask_feature_length": 10,
+  "mask_feature_min_masks": 0,
+  "mask_feature_prob": 0.0,
+  "mask_time_length": 10,
+  "mask_time_min_masks": 2,
+  "mask_time_prob": 0.05,
   "max_length": 448,
   "max_source_positions": 1500,
   "max_target_positions": 448,
   "pad_token_id": 50257,
   "scale_embedding": false,
   "torch_dtype": "float32",
+  "transformers_version": "4.27.0.dev0",
   "use_cache": true,
   "vocab_size": 51865
 }

eval_results.json CHANGED Viewed

@@ -1,9 +1,9 @@
 {
-    "epoch": 6.05,
-    "eval_cer": 123.45306368849984,
-    "eval_loss": 3.212874174118042,
-    "eval_runtime": 303.2793,
-    "eval_samples_per_second": 3.251,
-    "eval_steps_per_second": 1.626,
-    "eval_wer": 118.50381679389312
 }

 {
+    "epoch": 1.38,
+    "eval_cer": 48.17386054935105,
+    "eval_loss": 0.9147109389305115,
+    "eval_runtime": 340.4629,
+    "eval_samples_per_second": 2.896,
+    "eval_steps_per_second": 1.448,
+    "eval_wer": 48.06106870229008
 }

generation_config.json ADDED Viewed

	@@ -0,0 +1,225 @@

+{
+  "begin_suppress_tokens": [
+    220,
+    50257
+  ],
+  "bos_token_id": 50257,
+  "decoder_start_token_id": 50258,
+  "eos_token_id": 50257,
+  "forced_decoder_ids": [
+    [
+      1,
+      null
+    ],
+    [
+      2,
+      50359
+    ],
+    [
+      3,
+      50363
+    ]
+  ],
+  "is_multilingual": true,
+  "lang_to_id": {
+    "<|af|>": 50327,
+    "<|am|>": 50334,
+    "<|ar|>": 50272,
+    "<|as|>": 50350,
+    "<|az|>": 50304,
+    "<|ba|>": 50355,
+    "<|be|>": 50330,
+    "<|bg|>": 50292,
+    "<|bn|>": 50302,
+    "<|bo|>": 50347,
+    "<|br|>": 50309,
+    "<|bs|>": 50315,
+    "<|ca|>": 50270,
+    "<|cs|>": 50283,
+    "<|cy|>": 50297,
+    "<|da|>": 50285,
+    "<|de|>": 50261,
+    "<|el|>": 50281,
+    "<|en|>": 50259,
+    "<|es|>": 50262,
+    "<|et|>": 50307,
+    "<|eu|>": 50310,
+    "<|fa|>": 50300,
+    "<|fi|>": 50277,
+    "<|fo|>": 50338,
+    "<|fr|>": 50265,
+    "<|gl|>": 50319,
+    "<|gu|>": 50333,
+    "<|haw|>": 50352,
+    "<|ha|>": 50354,
+    "<|he|>": 50279,
+    "<|hi|>": 50276,
+    "<|hr|>": 50291,
+    "<|ht|>": 50339,
+    "<|hu|>": 50286,
+    "<|hy|>": 50312,
+    "<|id|>": 50275,
+    "<|is|>": 50311,
+    "<|it|>": 50274,
+    "<|ja|>": 50266,
+    "<|jw|>": 50356,
+    "<|ka|>": 50329,
+    "<|kk|>": 50316,
+    "<|km|>": 50323,
+    "<|kn|>": 50306,
+    "<|ko|>": 50264,
+    "<|la|>": 50294,
+    "<|lb|>": 50345,
+    "<|ln|>": 50353,
+    "<|lo|>": 50336,
+    "<|lt|>": 50293,
+    "<|lv|>": 50301,
+    "<|mg|>": 50349,
+    "<|mi|>": 50295,
+    "<|mk|>": 50308,
+    "<|ml|>": 50296,
+    "<|mn|>": 50314,
+    "<|mr|>": 50320,
+    "<|ms|>": 50282,
+    "<|mt|>": 50343,
+    "<|my|>": 50346,
+    "<|ne|>": 50313,
+    "<|nl|>": 50271,
+    "<|nn|>": 50342,
+    "<|no|>": 50288,
+    "<|oc|>": 50328,
+    "<|pa|>": 50321,
+    "<|pl|>": 50269,
+    "<|ps|>": 50340,
+    "<|pt|>": 50267,
+    "<|ro|>": 50284,
+    "<|ru|>": 50263,
+    "<|sa|>": 50344,
+    "<|sd|>": 50332,
+    "<|si|>": 50322,
+    "<|sk|>": 50298,
+    "<|sl|>": 50305,
+    "<|sn|>": 50324,
+    "<|so|>": 50326,
+    "<|sq|>": 50317,
+    "<|sr|>": 50303,
+    "<|su|>": 50357,
+    "<|sv|>": 50273,
+    "<|sw|>": 50318,
+    "<|ta|>": 50287,
+    "<|te|>": 50299,
+    "<|tg|>": 50331,
+    "<|th|>": 50289,
+    "<|tk|>": 50341,
+    "<|tl|>": 50348,
+    "<|tr|>": 50268,
+    "<|tt|>": 50351,
+    "<|uk|>": 50280,
+    "<|ur|>": 50290,
+    "<|uz|>": 50337,
+    "<|vi|>": 50278,
+    "<|yi|>": 50335,
+    "<|yo|>": 50325,
+    "<|zh|>": 50260
+  },
+  "max_initial_timestamp_index": 1,
+  "max_length": 448,
+  "no_timestamps_token_id": 50363,
+  "pad_token_id": 50257,
+  "return_timestamps": false,
+  "suppress_tokens": [
+    1,
+    2,
+    7,
+    8,
+    9,
+    10,
+    14,
+    25,
+    26,
+    27,
+    28,
+    29,
+    31,
+    58,
+    59,
+    60,
+    61,
+    62,
+    63,
+    90,
+    91,
+    92,
+    93,
+    359,
+    503,
+    522,
+    542,
+    873,
+    893,
+    902,
+    918,
+    922,
+    931,
+    1350,
+    1853,
+    1982,
+    2460,
+    2627,
+    3246,
+    3253,
+    3268,
+    3536,
+    3846,
+    3961,
+    4183,
+    4667,
+    6585,
+    6647,
+    7273,
+    9061,
+    9383,
+    10428,
+    10929,
+    11938,
+    12033,
+    12331,
+    12562,
+    13793,
+    14157,
+    14635,
+    15265,
+    15618,
+    16553,
+    16604,
+    18362,
+    18956,
+    20075,
+    21675,
+    22520,
+    26130,
+    26161,
+    26435,
+    28279,
+    29464,
+    31650,
+    32302,
+    32470,
+    36865,
+    42863,
+    47425,
+    49870,
+    50254,
+    50258,
+    50358,
+    50359,
+    50360,
+    50361,
+    50362
+  ],
+  "task_to_id": {
+    "transcribe": 50359,
+    "translate": 50358
+  },
+  "transformers_version": "4.27.0.dev0"
+}

merges.txt CHANGED Viewed

@@ -1,4 +1,4 @@
-#version: 0.2
 Ġ a
 Ġt h
 i n

+#version: 0.2 - Trained by `huggingface/tokenizers`
 Ġ a
 Ġt h
 i n

pred.txt CHANGED Viewed

@@ -1 +1 @@

- pred_str=['三頭無小 ', '三頭無無 ', '三頭無 ', '三頭人無人 ', '大人無 ', '三頭一人 ', '跤頭 ', '三頭無小 ', '大坑 ', '大湯 ', '三頭無 ', '跤頭 ', '三頭仔 ', '烏頭 ', '大人無 ', '跤頭 ', '大頭仔 ', '三頭仔 ', '大湲 ', '三人無無 ', '大人人人人人 ', '跤仔 ', '大人無無 ', '大坑 ', '三頭仔 ', '跤仔 ', '跤頭 ', '三頭人人人人生頂頂頂頂頂頂頂頂頂頂頂頂頂頂頂頂 ', '大頭 ', '大人無無 ', '大湲 ', '大頭仔 ', '徘 ', '大坑 ', '大角 ', '三頭無 ', '大人人人人人 ', '大湯 ', '大湲 ', '三頭頭 ', '大人無 ', '大大路 ', '大大路 ', '大坑 ', '三頭仔 ', '跤仔 ', '三頭無小 ', '大角仔 ', '跤頭 ', '大坵 ', '雞仔仔 ', '鵾仔 ', '大大路 ', '大人人人人 ', '三頭頭 ', '三頭仔 ', '三頭無 ', '大人無 ', '大磾 ', '跤仔 ', '跤頭 ', '三人無無人 ', '三頭仔 ', '跤仔 ', '大人無無 ', '大人人人人 ', '大火 ', '大磾 ', '大頭仔 ', '大磾 ', '大坑 ', '大湲 ', '大坵 ', '大人無 ', '大大路 ', '大湲 ', '大人無 ', '大人人人 ', '三頭仔 ', '大人無 ', '跤 ', '三頭無無 ', '大頭仔 ', '跤仔 ', '跤頭 ', '三頭無人 ', '鵭仔 ', '大頭仔 ', '烏頭 ', '三頭頭 ', '跤仔 ', '大磾 ', '三頭無無 ', '三頭無無 ', '三頭人人人 ', '三人人人一人生頂頂頂頂頂頂頂頂頂頂頂頂頂頂頂頂頂頂頂頂頂頂頂頂頂頂頂 ', '大頭仔 ', '跤頭 ', '大大路 ', '大頭 ', '跤仔 ', '跤頭 ', '跤仔 ', '鵭仔 ', '跤頭 ', '雞頭 ', '跤仔 ', '三頭頭 ', '跤仔 ', '大湯 ', '跤仔 ', '三頭人人一生一生 ', '三頭無小 ', '大爯 ', '大湲 ', '大人人人人 ', '三頭頭 ', '大人無 ', '三頭無無 ', '三頭頭 ', '大人無 ', '三人無人 ', '大大路 ', '三頭無無 ', '大頭仔 ', '大人無 ', '大坵 ', '大人無人 ', '大人無 ', '大人人人人 ', '大頭仔 ', '大湯 ', '跤頭 ', '大湯 ', '大坵 ', '跤仔 ', '大坑 ', '三頭人人生生一生 ', '三人人人人人生頂頂頂頂頂頂頂頂頂頂頂頂頂頂頂頂頂頂頂頂頂頂頂頂頂 ', '大人人人 ', '大大路 ', '三頭無無 ', '大坑 ', '大頭仔 ', '大人無 ', '大大路 ', '三頭無小 ', '三頭人一人一生一頂頂頂頂頂頂頂頂頂頂頂頂頂頂頂頂頂頂頂頂頂頂頂頂頂頂頂頂頂頂頂頂 ', '大人無 ', '大人無 ', '大人人人人 ', '大人無無 ', '大人人人人人 ', '大坵 ', '大坑 ', '大人無 ', '大大路 ', '大坵 ', '大人無無 ', '大人無無 ', '大人人人 ', '大人人人 ', '大坵 ', '大大路 ', '大湯 ', '三頭無小 ', '大人無無 ', '跤仔 ', '大坵仔 ', '三頭無人 ', '大人無 ', '三頭人一人生生 ', '大坵 ', '大人無 ', '三頭無無 ', '三頭頭 ', '跤仔 ', '三頭人人生生生 ', '三頭無小 ', '這人人人人人人人人伊人伊人伊人伊人伊人伊人 ', '大湯 ', '大人無 ', '大人無 ', '大火 ', '跤仔 ', '大湲 ', '跤仔 ', '大湲 ', '鵭仔 ', '徘 ', '大坑 ', '大坑 ', '大磾 ', '大湯 ', '大人無 ', '大人人人人 ', '大火 ', '三人無無 ', '三人人人人人人人人頂頂頂頂頂頂頂頂頂頂頂頂頂頂頂頂頂頂頂頂頂頂 ', '三頭人人一人 ', '三頭人人一人 ', '三頭人人一人 ', '三頭人人一生生 ', '三人無無 ', '三頭人人生生一生 ', '大人無無 ', '大人無無 ', '大人人人 ', '大人人人人 ', '三頭無人 ', '大頭無 ', '三頭人人人人 ', '大人無無 ', '三頭無 ', '三頭無無 ', '跤仔 ', '三頭無小 ', '跤頭 ', '三頭無小 ', '跤仔 ', '三頭無小 ', '跤頭 ', '三頭無小 ', '三頭無小 ', '跤仔 ', '跤仔 ', '三頭無小 ', '跤仔 ', '三頭無小 ', '三頭頭生 ', '三頭人一人生生生 ', '跤頭 ', '跤頭 ', '跤頭 ', '鵬 ', '雞頭 ', '大坵 ', '鵭 ', '跤仔 ', '跤頭 ', '大湲 ', '鵾仔 ', '跤仔 ', '大坑 ', '跤頭 ', '大坑 ', '大磾 ', '三頭無 ', '跤仔 ', '三頭無 ', '跤頭 ', '鵭 ', '大角仔 ', '大頭仔 ', '大角 ', '大火 ', '三頭無 ', '三頭頭 ', '跤仔 ', '三頭無小 ', '雞頭 ', '大人無 ', '大火仔 ', '大人無 ', '三頭仔 ', '大頭仔 ', '大頭仔 ', '大人無 ', '大人無無 ', '大坵 ', '三頭無無 ', '三頭無 ', '跤仔 ', '三頭人一人生生一生頂頂頂頂頂頂頂頂頂頂頂頂頂頂頂頂頂頂頂頂頂頂頂頂頂頂頂頂頂頂 ', '三頭頭 ', '大人無 ', '三頭頭 ', '三頭人人一生一生 ', '三頭人人 ', '大人無無 ', '三頭人一人生生生 ', '大人無無 ', '大頭無無 ', '三頭頭 ', '三頭一人生生生 ', '鵭仔 ', '三頭無無 ', '大湲 ', '大頭無無 ', '跤頭 ', '大頭 ', '大人無無 ', '三頭頭 ', '跤頭 ', '三頭仔 ', '跤仔 ', '三頭頭 ', '三頭頭 ', '大湲 ', '跤頭 ', '三頭頭 ', '三頭頭 ', '大大路 ', '跤頭 ', '跤仔 ', '三頭頭 ', '跤仔 ', '跤仔 ', '大角 ', '三頭仔 ', '跤頭 ', '跤頭 ', '跤頭 ', '跤仔 ', '鵬 ', '大湯 ', '大人無 ', '三頭人人生生生 ', '三頭無無 ', '三頭無小 ', '大人無 ', '大人無 ', '大角仔 ', '三頭無無 ', '三頭仔 ', '大頭 ', '三頭無無 ', '大湯 ', '這人人人人人人人人人人 ', '大湯 ', '大人無無 ', '大大路 ', '大大路 ', '大湲 ', '這人人人人人人人人人伊人 ', '跤仔 ', '大人無 ', '大人人人人 ', '三頭人人一人生頂頂頂頂頂頂頂頂頂頂頂頂 ', '大人無 ', '這人人人人人人人人人人 ', '大人無 ', '大人無無 ', '大湲 ', '大人無 ', '大頭仔 ', '大人無 ', '大坵 ', '大人無 ', '大人無 ', '大角 ', '大大路 ', '大大路 ', '大大路 ', '大湲 ', '大人人人人 ', '跤頭 ', '大人無 ', '大人無 ', '大人無 ', '大人人人 ', '大大路 ', '大大路 ', '大大路 ', '大人無 ', '大角仔 ', '三頭人人人人生頂頂頂頂頂頂頂頂頂頂頂頂頂頂 ', '大人人人人 ', '大人無 ', '大人無 ', '大人無 ', '大湯 ', '大人無無 ', '跤仔 ', '三頭無人 ', '三頭無小 ', '三頭無無 ', '大頭無無 ', '三頭無 ', '大人無 ', '大坑 ', '烏頭仔 ', '跤仔 ', '三頭人一人生生頂頂頂頂頂頂頂頂頂頂頂頂頂頂頂頂頂頂頂頂 ', '三頭仔 ', '跤頭 ', '大人無無 ', '雞頭 ', '跤頭 ', '三頭無小 ', '大人人人人人 ', '大人無 ', '大頭仔 ', '三頭人一人生生生一頂頂頂頂頂頂頂頂頂頂頂頂頂頂頂頂頂頂頂頂頂頂頂頂頂頂頂頂頂頂 ', '大人無 ', '跤仔 ', '三頭人一人生生生 ', '跤頭 ', '大角 ', '鵭仔 ', '大人無 ', '大人無 ', '跤仔 ', '大人無 ', '鵬 ', '大磾 ', '大大路 ', '大磾 ', '大坑 ', '大磾 ', '大人無 ', '大湯 ', '大坵 ', '大大路 ', '大大路 ', '大湯 ', '大湲 ', '大坵 ', '大火 ', '大大路 ', '大坑 ', '這人人人人人人人人人人 ', '大磾 ', '鵭仔 ', '大坑 ', '鵬 ', '大坑 ', '大人無 ', '大火 ', '大坑 ', '大坵 ', '大磾 ', '大磾 ', '大人無 ', '鵭 ', '大頭仔 ', '鵾仔 ', '跤仔 ', '大角 ', '三頭仔 ', '大湯 ', '三頭仔 ', '大人無 ', '大角 ', '大角 ', '三人人人人人頂頂頂頂頂頂頂頂頂頂頂頂頂頂頂頂頂頂頂 ', '大角仔 ', '大大路 ', '大大路 ', '三頭仔 ', '大人無無 ', '大角 ', '大角 ', '大角 ', '大坑 ', '大坵仔 ', '大人無 ', '大頭仔 ', '大湯 ', '大角 ', '大湲 ', '大人無 ', '大人無 ', '大人無 ', '大人無 ', '大大路 ', '大坵 ', '大湲 ', '大頭仔 ', '大頭仔 ', '大坵 ', '大頭 ', '大頭仔 ', '三人人人一人生頂頂頂頂頂頂頂頂頂頂頂頂頂頂頂頂頂頂 ', '大大路 ', '大人人人人 ', '大人無 ', '鵬 ', '大坑 ', '大人無 ', '大大路 ', '大大路 ', '大人人人人 ', '這人人人人人人人人人人伊人 ', '大火 ', '大坑 ', '大磾 ', '大坑 ', '大湯 ', '大湲 ', '大坵 ', '大湯 ', '大湲 ', '大磾 ', '大磾 ', '大磾 ', '大大路 ', '大坑 ', '大人無無 ', '大人無 ', '大坑 ', '大湲 ', '大湯 ', '大磾 ', '大坑 ', '大坑 ', '大大路 ', '大湯 ', '大磾 ', '大頭 ', '大湯 ', '大磾 ', '大磾 ', '大頭仔 ', '大人無 ', '大人人人人 ', '三頭無無 ', '大頭仔 ', '大頭仔 ', '三頭無小 ', '大頭 ', '大人無無 ', '三頭一人 ', '這人人人人人人人人人 ', '大頭仔 ', '三頭無無 ', '三頭無小 ', '大火 ', '三頭人人生生生 ', '三頭無人 ', '大人人人人 ', '三人人人人生頂頂頂頂頂頂頂頂頂頂頂頂頂頂頂頂頂頂頂頂頂頂頂頂頂頂頂頂頂 ', '三頭無無 ', '三頭一人生一生頂頂頂頂頂頂頂頂頂頂頂頂頂頂頂頂 ', '大坵 ', '大人無 ', '大人無 ', '大人人人人 ', '這人人人人人人人人人人人人伊人伊人伊人 ', '大人無 ', '三人無無 ', '三頭無小 ', '大人無無 ', '大火 ', '三頭仔 ', '三頭人人一人 ', '三頭一人生生頂頂頂頂頂頂頂頂頂頂頂頂 ', '三頭無人 ', '跤頭 ', '三頭仔 ', '大人無無 ', '這人人人人人人人人人人伊人 ', '三頭無 ', '大人無 ', '跤頭 ', '三頭仔 ', '大坵 ', '大坑 ', '大人無無 ', '大坵 ', '徘 ', '大坵 ', '大湲 ', '大湯 ', '大人無 ', '跤仔 ', '大大路 ', '大坵仔 ', '三頭人無人 ', '大坵 ', '大大路 ', '大人無 ', '大坵仔 ', '徘 ', '大大路 ', '大坑 ', '大坑 ', '大人無 ', '大湯 ', '大大路 ', '三頭無無 ', '大大路 ', '大湲 ', '大坑 ', '大人無 ', '大坑 ', '大大路 ', '大湲 ', '大坑 ', '大坵 ', '大坑 ', '大坵 ', '大大路 ', '大坵 ', '大火 ', '大人無 ', '大人人人人 ', '大湯 ', '大人無 ', '大大路 ', '大坵仔 ', '三頭無小 ', '跤仔 ', '大人無 ', '三頭無小 ', '三頭人人人生生一生頂頂頂頂頂頂頂頂頂頂頂頂頂頂頂頂頂 ', '大坑 ', '鵬 ', '大坑 ', '三頭無小 ', '大人無無 ', '三頭無 ', '跤仔 ', '跤頭 ', '大湲 ', '三頭仔 ', '跤仔 ', '跤仔 ', '跤頭 ', '三頭無無 ', '跤仔 ', '大人無 ', '大爯 ', '大湲 ', '大坑 ', '大人人人人 ', '三頭無小 ', '三頭無 ', '大坑 ', '跤頭 ', '三頭無 ', '大角 ', '大湯 ', '大人無 ', '大人無無 ', '大人無 ', '大坑 ', '大坵 ', '大人無 ', '三頭無 ', '三頭無 ', '大坑 ', '跤仔 ', '跤頭 ', '跤仔 ', '大頭仔 ', '三頭頭 ', '大頭 ', '跤仔 ', '三頭頭 ', '大火 ', '三頭無小 ', '跤頭 ', '三人無無無 ', '大人無人 ', '大人無 ', '鵬 ', '跤仔 ', '大火 ', '鵭仔 ', '大頭仔 ', '跤頭 ', '三頭人人一生一生頂頂頂頂頂頂頂頂頂頂頂頂頂頂頂頂頂頂頂 ', '大人無無 ', '大頭仔 ', '鵭仔 ', '大人無 ', '大人無無 ', '雞頭 ', '鵾仔 ', '大角 ', '大磾 ', '三頭仔 ', '跤頭 ', '三頭仔 ', '大角仔 ', '三頭仔 ', '大人無 ', '大人無 ', '鵭仔 ', '跤仔 ', '三頭無 ', '烏頭 ', '三頭仔 ', '三頭人人一人 ', '三頭仔 ', '跤頭 ', '跤仔 ', '三頭無無 ', '三頭人無人 ', '大人無 ', '跤仔 ', '三頭無小 ', '大頭仔 ', '三頭無 ', '大頭仔 ', '三頭人一人生生生生頂頂頂頂頂頂頂頂頂頂頂頂頂頂頂頂頂頂頂頂頂頂頂頂頂頂 ', '大頭仔 ', '三頭無無 ', '跤仔 ', '大火 ', '大磾 ', '大磾 ', '大人無 ', '鵬 ', '跤仔 ', '跤頭 ', '鵭仔 ', '跤仔 ', '大人無 ', '大人無 ', '大頭路 ', '大人無 ', '雞頭 ', '大坑 ', '大坑 ', '大人無無 ', '大人無 ', '三頭無小 ', '徘 ', '大坑 ', '大磾 ', '鵭仔 ', '大人無無 ', '大湲 ', '大火 ', '大人無 ', '大人人人 ', '大大路 ', '大坑 ', '大角仔 ', '大人人人 ', '大磾 ', '大爯 ', '大磾 ', '大人人人人 ', '大磾 ', '大人無 ', '大人無無 ', '大湯 ', '大磾 ', '大火 ', '大湲 ', '大湲 ', '大湲 ', '大爯 ', '大坑 ', '徘 ', '徘 ', '大坑 ', '大坑 ', '大磾 ', '徘 ', '大人無 ', '大人無 ', '大坑 ', '大坵 ', '徘 ', '大坑 ', '大坵 ', '大大路 ', '鵬 ', '跤仔 ', '徘 ', '大大路 ', '大大路 ', '大人無 ', '大坑 ', '大坑 ', '大人無 ', '大人無 ', '大湲 ', '大大路 ', '三頭無無 ', '三頭無小 ', '跤仔 ', '大坑 ', '大人無 ', '三頭人一人生生生 ', '大人無無 ', '三頭無 ', '大頭仔 ', '大湲 ', '大人無無 ', '大人無 ', '大人無 ', '三頭無 ', '大人人人人 ', '三頭一人 ', '三頭人一人生生 ', '跤仔 ', '大人無 ', '三頭一人 ', '三人無人 ', '跤仔 ', '大人無無 ', '三頭人人一生生生 ', '大頭仔 ', '三頭頭 ', '大頭 ', '大人人人人 ', '大湲 ', '大人無 ', '大角 ', '大人人人 ', '三頭無小 ', '三頭頭生 ', '大人無 ', '大人無 ', '大人人人人 ', '大人人人 ', '三頭仔 ', '大人人人 ', '三頭人人一人 ', '三頭仔 ', '三頭無小 ', '大頭 ', '大人無無 ', '大坵 ', '三頭無人 ', '大磾 ', '大火仔 ', '大湲 ', '跤頭 ', '跤頭 ', '三頭頭 ', '三頭無無 ', '三頭無無 ', '三頭無生 ', '三頭仔 ', '三頭人一人生生頂頂頂頂頂頂頂頂頂頂頂頂頂頂頂頂頂頂頂頂頂頂頂頂頂頂頂 ', '三頭仔 ', '大人無 ', '大大路 ', '大人無 ', '烏頭仔 ', '大人無無 ', '大人無 ', '大人無 ', '大人無 ', '大湯 ', '跤頭 ', '大頭無 ', '三人人人人人人人人人人一頂頂頂頂頂頂頂頂頂頂頂頂頂頂頂頂頂頂頂頂頂頂頂 ', '三人無人 ', '大頭 ', '大火 ', '跤頭 ', '大人無 ', '鵾仔 ', '跤仔 ', '三頭無無 ', '大人無無 ', '大大路 ', '三頭仔 ', '大湲 ', '大湯 ', '大湲 ', '大磾 ', '跤仔 ', '大人無 ', '大火仔 ', '大頭仔 ', '三頭頭 ', '鵬 ', '大人無 ', '大人無 ', '大人無 ', '大人無 ', '大湲 ', '跤頭 ', '跤頭 ', '三頭無無 ', '大大路 ', '大坵 ', '鵭仔 ', '大人無無 ', '大人無 ', '三頭無小 ', '大火 ', '大大路 ', '大頭 ', '鵭 ', '大大路 ', '大磾 ', '跤仔 ', '大大路 ', '大人無 ', '大人無 ', '大火 ', '大人無 ', '大頭仔 ', '大湯 ', '大磾 ', '大磾 ', '大磾 ', '大湲 ', '大湯 ', '大湲 ', '大人無 ', '大湯 ', '大人無 ', '大湯 ', '大人人人人人 ', '大磾 ', '大湲 ', '大人無 ', '大磾 ', '大磾 ', '大人無 ', '大湲 ', '大湲 ', '大人無 ', '大人無 ', '大大路 ', '大人人人 ', '大人無 ', '大人無無 ', '大人無無 ', '大湲 ', '大人人人人人 ', '大人無 ', '大大路 ', '大人無 ', '大人人人人 ', '大人無 ', '大人無 ', '大人無 ', '大人無 ', '大人人人人 ', '大人無無 ', '大湯 ', '大湯 ', '大人無 ', '大人無 ', '大大路 ', '大人無 ', '三頭無 ', '大湲 ', '大火 ', '大頭仔 ', '大人無 ', '大人無 ', '三頭人人一人生頂頂頂頂頂頂頂頂頂頂頂頂頂頂頂頂頂頂頂頂頂 ', '大湲 ', '大湲 ', '大湲 ', '大湯 ', '大坑 ', '跤仔 ', '大人無 ', '大人無 ', '大磾 ', '大大路 ', '鵭仔 ', '大磾 ', '跤仔 ', '大湲 ', '三頭無無 ', '大大路 ', '大人人人人 ', '大人無 ', '大人無 ', '大火 ', '三頭人人 ', '大湲 ', '大大路 ', '大人無 ', '大頭仔 ', '大坵 ', '大坑 ', '大人無無 ', '大湲 ', '大人無 ', '大人無 ', '大湯 ', '大人無 ', '大火 ', '大湲 ', '大人無 ', '大磾 ', '大人無 ', '大人無 ', '大人無 ', '大湲 ', '大湯 ']

+ pred_str=['心理性別 ', '火燒跤村 ', '草屬 ', '十三項 ', '水里鄉 ', '這代有一個大商場 ', '聯合國 ', '菜頭人仔 ', '若洲 ', '肉羹 ', '新興區 ', '太古幣 ', '麵冬 ', '喙仔 ', '洗蚵飯 ', '新粹 ', '燒包仔 ', '空頭 ', '黃坤 ', '落水溪出代誌 ', '粗細 ', '愛害 ', '九年一貫 ', '精靈 ', '三寶 ', '速速 ', '蔗菜咧擲 ', '眼目袂消看關係 ', '鎮烏 ', '兩个人已经講和矣 ', '濁水 ', '細空喙 ', '掠著筋頭 ', '復興館 ', '戒律 ', '草人坑 ', '祝你生理快樂 ', '坑潭 ', '出水 ', '皮皮串 ', '囝仔羹 ', '菝露 ', '臺灣人 ', '落南 ', '原理坑 ', '坑仔內 ', '起冊到地 ', '戶錢 ', '紅糖 ', '青大湖 ', '柴柴橋 ', '柴球 ', '大胃王子 ', '大橂埕 ', '頭類 ', '桑餅 ', '喙暗籠 ', '龍門 ', '時局 ', '熟士 ', '線尾 ', '無話講家老 ', '下暗磅 ', '一人 ', '一定看袂著 ', '看一下影生一下囝 ', '倒勢 ', '直行 ', '鬼進壽仙 ', '肚肚 ', '變天 ', '一工到鴨 ', '粉盤 ', '有幾項菜 ', '岩埔 ', '鑼角區 ', '過路山 ', '戇買戇人行 ', '粗食 ', '這咧外濟錢 ', '斜仔 ', '看甲清清粗粗 ', '妄生 ', '發聲 ', '白囡仔 ', '心火動火 ', '番鼎仔 ', '流窟窟 ', '大溪火 ', '名聲真頭 ', '閃烘肉 ', '骨 ', '龜大 ', '水溪皮仔 ', '夯月糕 ', '翹無三文門著想欲做塗公 ', '尪姨仔 ', '臺北橋 ', '雲林縣 ', '臺中港 ', '水里 ', '臺南市 ', '七度 ', '做公竹 ', '臺川鄉 ', '燒肉肉 ', '筊 ', '漢草真好 ', '殘影 ', '按呢生 ', '海外 ', '無味可拄著潤骨 ', '奇鄉小細 ', '復 ', '頭尾 ', '牛頓市場 ', '政商 ', '吐著咧 ', '動物性格質 ', '實問精 ', '悠悠殼 ', '鑼鑼霧小睏 ', '乞頭鄉 ', '聖義臺安路口 ', '三三八八 ', '你話毋通予白講 ', '白寮鄉 ', '你莫著中午肚臍 ', '新安路 ', '金錢毋是萬能的 ', '飛苓雅菜 ', '富寮里 ', '你是誰 ', '番烏 ', '神經 ', '合同 ', '挺挺 ', '販賣殼 ', '天氣冷啦你就愛街踏一件衫 ', '我到底看著啥物 ', '手環 ', '古不日三彰 ', '廳價 ', '透睛仔 ', '傲求術 ', '隔音壁 ', '拍桶頭 ', '揮火滿天篷結智者行人 ', '東北貴方 ', '武情機 ', '做 �� 無小力 ', '景空白虹 ', '這妖仔摸落去真緊張的體育仔 ', '毛髏仔 ', '割喙血 ', '陽明山 ', '加鼻灌 ', '囡仔食 ', '佇對路症一點情 ', '伊歸心美虎大學 ', '東湖國中 ', '中原公園 ', '半路店 ', '冰冷坑 ', '東石鄉 ', '餅酥的 ', '無中生化 ', '馬藏角 ', '演藝伎 ', '厝邊頭背 ', '波灣粿 ', '仁愛烘酒路口 ', '千頁 ', '七夕肚 ', '大極大理 ', '烘仔頭鏡 ', '飼 ', '咱做代誌的時尊 ', '包水餃 ', '海鳥毋知飛海柴破袂開 ', '溫宗中 ', '健康中生 ', '印度尼西亞 ', '生理族 ', '獅仔頭 ', '大肚路 ', '市仔 ', '落野區 ', '學姐 ', '放人山 ', '虎尾直 ', '下腹肚 ', '興村鎮 ', '燒火球 ', '化極 ', '我三攏穿上大領 ', '溜下來 ', '半條命 ', '生活習慣 ', '電影機 ', '菜脯蛋 ', '塗塗塗 ', '鐵甲牛 ', '風火頭 ', '現流仔 ', '培了工 ', '空港 ', '較大 ', '公政橋 ', '巴結 ', '古汁 ', '有拜無積穏 ', '這時 ', '直白 ', '下災 ', '較早睏 ', '寮寮臺巡 ', '公司 ', '懸枯緊緊 ', '坑仔 ', '林線香菇 ', '山羊 ', '大頭九寶 ', '塗辦人道 ', '對 ', '內沿 ', '阮會香櫞 ', '地目 ', '真有喙水 ', '模特師 ', '相相認你就是林朋友 ', '青山溫 ', '鶴曲蘭 ', '金冬 ', '變啦 ', '單欲節 ', '狗尾金 ', '牛坑 ', '過冬跤 ', '頭辦坑坑 ', '無聽著 ', '礦機仔 ', '厝影 ', '江州 ', '青天街 ', '東方 ', '問題 ', '桃源聖別 ', '故厝 ', '妖嬌 ', '復國小 ', '點磕 ', '光臺 ', '目虛股仔 ', '立斗 ', '永環 ', '你是我我 ', '達著歹隊步 ', '僥候 ', '骨髏公的無船上 ', '加辣片 ', '如來如嚴重 ', '一個一個 ', '陳公大橋 ', '上大大師所 ', '蒜仔虯公 ', '骨銅跤 ', '復興戶 ', '講啥芋仔漢字 ', '毋行 ', '親密閨讀著童 ', '你使毋知哉臭 ', '斗南 ', '狗睏鼠無趖 ', '水粿仔 ', '門狀 ', '五手空 ', '心無目無厄 ', '敢食氣溜 ', '阿婆婆 ', '機動戰士 ', '電動 ', '起直行 ', '跤先股 ', '所羅門群島 ', '彰界 ', '我愛大股 ', '高頂 ', '甜鹹 ', '上界 ', '是小 ', '落億不折 ', '跤倉門 ', '猴仔疊豬仔 ', '風神尪仔 ', '山 ', '生甲袂蠻 ', '九月趖 ', '橫刀仔 ', '中央大學 ', '石門水庫 ', '亂史上 ', '百貨店 ', '酸素肝 ', '甲界 ', '洗衫飯 ', '天臺 ', '數位拍 ', '後褲 ', '鈍刺火 ', '頭走位 ', '夭壽貴 ', '好所在 ', '兩市 ', '鐵理 ', '現辦 ', '九點位 ', '伊生做這麼臭老 ', '摩星仔石鉎 ', '一甘甘仔 ', '龍公 ', '隔十五件 ', '提著 ', '心心酒店 ', '過時抵時 ', '較平咧 ', '桃生大罐 ', '戮區杯仔 ', '行佇這袂著愛工薪 ', '隔花仔 ', '抹衫面仔 ', '雨傘鵰 ', '石板 ', '虎魚 ', '落雨了塗跤鼻公擔落落 ', '較倉 ', '家人市 ', '你欲行對度位去 ', '敦化和平路口 ', '公仔喙 ', '阮囝今年二十出頭矣 ', '新港域 ', '愛袂叫母 ', '一分 ', '問問仔戶 ', '元安 ', '牛丼 ', '袂看得 ', '立公立直 ', '牛山肚 ', '過度 ', '十做水 ', '麥仔跤 ', '包大 ', '原地 ', '三重三重 ', '林口 ', '必定 ', '初度 ', '揣轉來 ', '機中站 ', '樹仔代 ', '心股 ', '小大 ', '頂底寮 ', '佇人 ', '你甲包仔提起哈少一下 ', '病風景病風君 ', '頂公館 ', '線仙 ', '演技界 ', '花粉 ', '毋情毋怨 ', '較緊 ', '有也好無也好 ', '草生 ', '講喙仔龍 ', '怪奇果 ', '國土家房 ', '新羅 ', '貴妻 ', '豆豆仔 ', '拍寶 ', '物件若予伊提過手 ', '電腦霓 ', '意界 ', '刺尾風 ', '烏內 ', '提灶 ', '無搭無散 ', '我欲遵佇落山 ', '倒擁 ', '聽無 ', '彼个頭家娘對人客食好咧 ', '大細練 ', '無葉 ', '無二五你的出國讀冊 ', '石烘 ', '嘉義 ', '後寮 ', '我攏有看著 ', '鏡仁路 ', '歹天 ', '三跤步一坎點 ', '臘肉 ', '句裡作 ', '下午中生 ', '林棍 ', '無直結 ', '公文 ', '十喙九跤倉 ', '頭仔片 ', '掠錢 ', '無空鳥 ', '蚵仔湯仔 ', '隔屎面 ', '白對銅 ', '膨一枝 ', '無一步 ', '老口講出來 ', '斗頂 ', '膨底做甲流懸膨跤成甲流懸 ', '烏豆汁 ', '拍屎辦 ', '桶邊 ', '焦巴 ', '簡訊 ', '一下才一下 ', '筋的酸 ', '洋仔話 ', '拍數 ', '直落挖 ', '總舖 ', '三里無見久留留 ', '學官 ', '手心動 ', '電子 ', '暗空 ', '散食 ', '速加車 ', '妖意 ', '燒雞龜 ', '�� 鉤 ', '寮溝 ', '暗頭仔 ', '濟囝濟白白濟媳婦濟體育 ', '溝球 ', '無睹好 ', '戳頭鵰 ', '金環實力 ', '觀音大師 ', '臺準 ', '欲按嘛 ', '後路 ', '結尾 ', '一萬兩仔 ', '提款片 ', '無你的代 ', '天公祖 ', '篡港 ', '麗茶 ', '馬鈴薯條 ', '做塗水的 ', '林青味 ', '流明潭 ', '官包遊 ', '糖醋 ', '二戰 ', '拍鐵屎 ', '慢一跤步 ', '提皮的 ', '下禮拜 ', '死亡之鴨 ', '你毋通囥佇身外內 ', '臭菇仔 ', '歎一空食三冬 ', '我欲學臺語 ', '臺 ', '目角肚 ', '雨水暗 ', '貴州省 ', '敦華北路 ', '你看起來是才拿仔水 ', '情的食妄妄的食天公 ', '轉去 ', '那不是 ', '幾個 ', '防長 ', '港墘 ', '二寵仔 ', '無鐵仔 ', '原型力 ', '麥坑 ', '六點 ', '市中 ', '樹母苓 ', '魚仔羹 ', '廣寮區 ', '柑花噴天 ', '大好逆人 ', '行政院 ', '食市 ', '九底厝 ', '寧星 ', '月光 ', '種水氣 ', '奇幼 ', '天公市 ', '懸 ', '愛大理 ', '我薑尾袂負矣 ', '天方 ', '有連線 ', '衝 ', '中國話 ', '拔香對拜 ', '四跤草 ', '挺風 ', '觀覽 ', '結上路 ', '彈跳 ', '焦工算 ', '時過外大平 ', '有這款的代誌我攏毋知 ', '磨豆仔機 ', '國聖夜 ', '林口區 ', '菝 ', '苦蒜明薯 ', '代誌圖案 ', '半生半疑 ', '點甜點甜土坑兄味事 ', '你有工工 ', '仁堂無洗手的朋友 ', '烏日 ', '微風 ', '水旋 ', '懸欠 ', '連半个人影也無 ', '調褲 ', '塗水公 ', '無沙沙 ', '微星船 ', '欲玩矣 ', '韌種 ', '頂頭目睭咧念字 ', '伊有誠濟內筍 ', '工仔喙國小 ', '濁國速公 ', '落尾 ', '平方電腦 ', '問蠓仔烏鑼鼓塗麥擔 ', '仁君水 ', '漁人副按一下位 ', '鹹酥雞 ', '天然蚵仔 ', '熟悾 ', '茨杜 ', '聖誕老阿公 ', '天頂白茫茫 ', '飯煎 ', '南巡 ', '吾人事 ', '有溫 ', '包山包海 ', '點空區 ', '洗衫精 ', '哈勁鎮商 ', '開喙罨粒粒草 ', '焦水器 ', '老舊家 ', '速加路亞 ', '數位仔聯名 ', '臭水 ', '屬暗君仔 ', '飛行機 ', '冰粿 ', '圍空天下不亂 ', '錢票 ', '龜仔樹 ', '溶角三角 ', '半夜 ', '思維 ', '衝勇 ', '敢有問題 ', '遮涼 ', '月磯忝竹 ', '做塗水 ', '怪奇 ', '班景 ', '狡官 ', '歹牙 ', '中山路 ', '橫仔溝 ', '粉圓仔 ', '菊菊摸摸 ', '毋知佇看個時 ', '妖冬 ', '激光雨 ', '海塗仔 ', '溝頭果 ', '熟士娘上 ', '面珠 ', '底葉卵 ', '無時焦電 ', '斜人一口勁人一道 ', '鐵磯 ', '銅杯仔 ', '二刀 ', '復生豆子隻 ', '人講對人講 ', '金剛鑽石 ', '做肉餅 ', '小寶喙 ', '櫳哥魚 ', '比頭村仔 ', '一筆錢 ', '菝菜 ', '溝仔密情 ', '一枝草一點路 ', '早市 ', '大頭母人 ', '包的 ', '上禮拜 ', '牛厝仔 ', '人行貴貴行人 ', '無記地情 ', '日落山 ', '憨鄉 ', '七塗跤 ', '袂死心 ', '解濟 ', '做遊戀 ', '有聽著無 ', '大姨田 ', '斗人仔本 ', '固固 ', '影人 ', '勥查戶 ', '暗端食生 ', '草薑流角 ', '挺快 ', '首寶 ', '薄餅 ', '手著 ', '亂無 ', '無情菜 ', '內街 ', '烏樹 ', '小酸臭 ', '菊潤 ', '蠓公廁貴 ', '山炭 ', '人聲欠貨 ', '天公無帶著 ', '望遠鄉 ', '食風 ', '粉刺 ', '皮車 ', '大姊 ', '公館路 ', '碧潭路 ', '臺北歐洲校校小學部 ', '倉裡不分 ', '四分仔 ', '頭殼的 ', '南京城 ', '應隨機 ', '破珠溪 ', '池上 ', '跤車 ', '懸胎 ', '苦袂當 ', '單從工 ', '喙驚和 ', '斜浮 ', '加攏城 ', '頂底蚵 ', '張張假口 ', '社群 ', '用頂厝 ', '風駝公園 ', '金烏鎮 ', '中校復興 ', '民族區溫度口 ', '松信路 ', '飼香粉 ', '溫州 ', '臺北聯絡線 ', '有夠咧加速 ', '伊加我一半 ', '酸仔 ', '記者鄉 ', '驚柑仔囝 ', '四跤水仙 ', '暖暖區 ', '用齒面仔洗喙時愛上律氣膏 ', '新社 ', '母親力顧予伊好 ', '十包 ', '部位 ', '埔路 ', '冰糖 ', '復興橋 ', '阿拉 ', '蒜蠻 ', '員林市 ', '批錢 ', '妄大志 ', '麵腸仁 ', '景興路 ', '海蛤仔頭 ', '一月半月 ', '失聯制 ', '後區 ', '隨風 ', '蘭斯貓頭鳥 ', '茅露開 ', '子宮喙暗 ', '高姊太 ', '假說 ', '不知 ', '院頭區 ', '上海茂霞 ', '意識面 ', '光風 ', '月耳池中 ', '彼人災我無 ', '紅豆仔蜂 ', '頂河 ', '商機 ', '三兩人講四斤話 ', '金貨 ', '一組 ', '軟褲 ', '珍珠看著鳥市屎 ', '天津 ', '角運動量 ', '�� 問著恐怖 ', '貧極競 ', '羅連 ', '真聽話 ', '溶炭煎 ', '火星岩 ', '可睏員 ', '烏上 ', '椅頭風 ', '蔭喙 ', '查緯 ', '較通 ', '番薯片 ', '頂去 ', '卵珠 ', '歡喜感源 ', '大海無驚大水 ', '懷矣 ', '跤手電 ', '欲 ', '煎巴醬 ', '徛肚胿 ', '查某人 ', '韓子球 ', '膨了臭 ', '大火 ', '火炬母 ', '英國話 ', '明天顧問 ', '老機 ', '看有 ', '影差天火度 ', '下文外童 ', '大路二 ', '內 ', '金風山 ', '死袂見笑 ', '蟮子 ', '焦化 ', '馬鈴薯片 ', '你頭拍甲倉仔顧無客 ', '囊物褲粉 ', '擋水機 ', '肚肚胿仔 ', '陂南 ', '袂當敢若我看著 ', '有困山 ', '大市好勢仔蠢 ', '水氣 ', '景文街 ', '樹天樹地 ', '今年是一個好年冬 ', '事搭 ', '雨傘頭 ', '三講四無著 ', '五所分離 ', '反插 ', '大日本帝國 ', '烘蓋芒霜 ', '臘蚵 ', '吞兩 ', '莫計 ', '經濟做代誌較要緊 ', '你講啥 ', '景水 ', '位極 ', '身體健康 ', '跤手十足 ', '延珠學 ', '下年 ', '大路大股 ', '伊是一下大美人 ', '解溶 ', '放棄 ', '你陪我去彼邊好無 ', '網路用較刁咧 ', '樹屎 ', '孤單一生 ', '加滴 ', '市長袂中 ', '有手仔 ', '孤烘毒市 ', '裡面 ', '自目 ', '役仔 ', '燒狗 ', '冷喙 ', '創辦人 ', '水灝蠟 ', '飼養動物 ', '目睭白金 ', '撲囡仔 ', '我予滾水蟲一下膨拍著聽的 ', '聖杯 ', '為人 ', '過濟 ', '三二九 ', '飯焦 ', '天車 ', '圍篤 ', '目睭久 ', '白紅 ', '電蜜 ', '鹽酸臭 ', '粉仔韓市 ', '你一次來伊攏居好寮的 ', '來喲今來看喲 ', '犯改 ', '總事記 ', '破柴 ', '掠開 ', '鳳山 ', '老厝 ', '開船母昌 ', '金美國小 ', '直白市 ', '橋頭區 ', '南州 ', '好味鎮 ', '長力 ', '西區 ', '國泰 ', '水腸流 ', '中庸路 ', '南福著 ', '金華大學 ', '西角 ', '樹林區 ', '紙片變膠 ', '中山橋 ', '彰化銀行 ', '吳國昌 ', '喙風橋 ', '花仔菜 ', '中情國小 ', '這樣就好 ', '老去矣 ', '車頂 ', '南洋街 ', '匈羅 ', '松壽路口 ', '中坑 ', '不屬自客 ', '瑞源 ', '後役 ', '大腸鼎 ', '筊仔弓 ', '再拜 ', '新營區 ', '政二幼局 ', '寧夏路 ', '動慢 ', '大好年 ', '仁愛鄉 ', '快龍 ', '辦 ', '洗面 ', '食肌 ', '調官機 ', '仙獅娘 ', '拍鐵寮 ', '路營區 ', '韌尾 ', '外鄉客 ', '聖字區 ', '你阿母間災仔你在遮發貨物 ', '敢安呢 ', '到草原 ', '望星橋 ', '市府 ', '實字架 ', '文昌橋 ', '有影緊無 ', '林內鄉 ', '紅目樹屎 ', '囡仔人目看有 ', '菁大 ', '文古揣 ', '歹喙道 ', '查某人 ', '滷菜頭 ', '跳過 ', '鴨母帶唱旋 ', '不貪不自 ', '客行 ', '下腹腓 ', '天安門公場 ', '豆磨 ', '溫寮戶 ', '初花 ', '雞籠蜂 ', '暗明網線娘 ', '袂曉公話 ', '腰床 ', '順勢 ', '三攏無 ', '鐵欲膠 ', '推渡 ', '卒久無看的 ', '起家補貨 ', '臺東 ', '不見天 ', '麥花蜜 ', '著蝦仔腸 ', '苦寮仔 ', '五肌筋頭好隔腿的突肌嘛痛 ', '溶毛位 ', '厝尾頂 ', '橫樹人 ', '大坵 ', '殼羹 ', '客戶 ', '按摩師 ', '芋仔糊 ', '海口 ', '阿里不搭 ', '冰島 ', '摺 ', '屎啦 ', '臭屎青 ', '網紗尿尻倉 ', '策伐 ', '庸桑歸桑 ', '一切攏是假的 ', '小山崙仔 ', '煎仔糊 ', '青眉仔目鏡 ', '親愛的 ', '臺大大學 ', '欲去到位 ', '坑頭鴨 ', '面流 ', '揣著 ', '嬌童小道 ', '門東仔 ', '寧角龍 ', '狡頭不勇 ', '桃園 ', '我欲去便所 ', '彼當時 ', '片雷公 ', '代誌問號 ', '症頭 ', '我思緘 ', '材料庫學 ', '甕底水溪 ', '貨錢 ', '睏袂去 ']

preprocessor_config.json CHANGED Viewed

The diff for this file is too large to render. See raw diff

pytorch_model.bin CHANGED Viewed

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:9c1e1cf1e0f592d5e4a409fb4bad5c85a3999010e244b51f776a67a2fc6c0616
 size 6173655480

 version https://git-lfs.github.com/spec/v1
+oid sha256:4ef4e1637cc218a82689705eefba3dfb4c61dd4754c6462e1492eb4c9dee2738
 size 6173655480

tokenizer.json ADDED Viewed

The diff for this file is too large to render. See raw diff

tokenizer_config.json CHANGED Viewed

@@ -19,7 +19,6 @@
   },
   "errors": "replace",
   "model_max_length": 1024,
-  "name_or_path": "openai/whisper-large-v2",
   "pad_token": null,
   "processor_class": "WhisperProcessor",
   "return_attention_mask": false,

   },
   "errors": "replace",
   "model_max_length": 1024,
   "pad_token": null,
   "processor_class": "WhisperProcessor",
   "return_attention_mask": false,

train_results.json CHANGED Viewed

@@ -1,7 +1,7 @@
 {
-    "epoch": 6.05,
-    "train_loss": 3.0522214595794677,
-    "train_runtime": 5458.8622,
-    "train_samples_per_second": 1.832,
-    "train_steps_per_second": 0.916
 }

 {
+    "epoch": 1.38,
+    "train_loss": 0.91982373046875,
+    "train_runtime": 1347.6358,
+    "train_samples_per_second": 1.484,
+    "train_steps_per_second": 0.742
 }

trainer_state.json CHANGED Viewed

@@ -1,1275 +1,41 @@
 {
   "best_metric": null,
   "best_model_checkpoint": null,
-  "epoch": 6.0532,
-  "global_step": 5000,
   "is_hyper_param_search": false,
   "is_local_process_zero": true,
   "is_world_process_zero": true,
   "log_history": [
     {
-      "epoch": 0.01,
-      "learning_rate": 4.2000000000000006e-07,
-      "loss": 2.3259,
-      "step": 25
-    },
-    {
-      "epoch": 0.01,
-      "learning_rate": 9.000000000000001e-07,
-      "loss": 1.7615,
-      "step": 50
-    },
-    {
-      "epoch": 0.01,
-      "learning_rate": 1.3800000000000001e-06,
-      "loss": 1.2375,
-      "step": 75
-    },
-    {
-      "epoch": 0.02,
-      "learning_rate": 1.8800000000000002e-06,
-      "loss": 1.3112,
-      "step": 100
-    },
-    {
-      "epoch": 0.03,
-      "learning_rate": 2.38e-06,
-      "loss": 1.3986,
-      "step": 125
-    },
-    {
-      "epoch": 0.03,
-      "learning_rate": 2.88e-06,
-      "loss": 1.6643,
-      "step": 150
-    },
-    {
-      "epoch": 0.04,
-      "learning_rate": 3.3800000000000007e-06,
-      "loss": 1.7064,
-      "step": 175
-    },
-    {
-      "epoch": 0.04,
-      "learning_rate": 3.88e-06,
-      "loss": 1.7026,
-      "step": 200
-    },
-    {
-      "epoch": 0.04,
-      "learning_rate": 4.38e-06,
-      "loss": 2.0459,
-      "step": 225
-    },
-    {
-      "epoch": 0.05,
-      "learning_rate": 4.880000000000001e-06,
-      "loss": 2.3189,
-      "step": 250
-    },
-    {
-      "epoch": 0.06,
-      "learning_rate": 5.380000000000001e-06,
-      "loss": 2.647,
-      "step": 275
-    },
-    {
-      "epoch": 0.06,
-      "learning_rate": 5.8800000000000005e-06,
-      "loss": 2.4019,
-      "step": 300
-    },
-    {
-      "epoch": 0.07,
-      "learning_rate": 6.380000000000001e-06,
-      "loss": 2.9076,
-      "step": 325
-    },
-    {
-      "epoch": 0.07,
-      "learning_rate": 6.860000000000001e-06,
-      "loss": 4.1248,
-      "step": 350
-    },
-    {
-      "epoch": 0.07,
-      "learning_rate": 7.360000000000001e-06,
-      "loss": 2.9751,
-      "step": 375
-    },
-    {
-      "epoch": 0.08,
-      "learning_rate": 7.860000000000001e-06,
-      "loss": 3.1863,
-      "step": 400
-    },
-    {
-      "epoch": 0.09,
-      "learning_rate": 8.36e-06,
-      "loss": 3.1017,
-      "step": 425
-    },
-    {
-      "epoch": 0.09,
-      "learning_rate": 8.860000000000002e-06,
-      "loss": 3.3757,
-      "step": 450
-    },
-    {
-      "epoch": 0.1,
-      "learning_rate": 9.360000000000002e-06,
-      "loss": 4.871,
-      "step": 475
-    },
-    {
-      "epoch": 0.1,
-      "learning_rate": 9.86e-06,
-      "loss": 7.6053,
-      "step": 500
-    },
-    {
-      "epoch": 0.1,
-      "learning_rate": 9.960000000000001e-06,
-      "loss": 4.4155,
-      "step": 525
-    },
-    {
-      "epoch": 0.11,
-      "learning_rate": 9.904444444444445e-06,
-      "loss": 4.5183,
-      "step": 550
-    },
-    {
-      "epoch": 0.12,
-      "learning_rate": 9.84888888888889e-06,
-      "loss": 4.1442,
-      "step": 575
-    },
-    {
-      "epoch": 0.12,
-      "learning_rate": 9.793333333333333e-06,
-      "loss": 4.1241,
-      "step": 600
-    },
-    {
-      "epoch": 0.12,
-      "learning_rate": 9.737777777777779e-06,
-      "loss": 3.9046,
-      "step": 625
-    },
-    {
-      "epoch": 0.13,
-      "learning_rate": 9.682222222222223e-06,
-      "loss": 3.8142,
-      "step": 650
-    },
-    {
-      "epoch": 0.14,
-      "learning_rate": 9.626666666666667e-06,
-      "loss": 3.6618,
-      "step": 675
-    },
-    {
-      "epoch": 0.14,
-      "learning_rate": 9.571111111111113e-06,
-      "loss": 3.6848,
-      "step": 700
-    },
-    {
-      "epoch": 0.14,
-      "learning_rate": 9.515555555555557e-06,
-      "loss": 3.6017,
-      "step": 725
-    },
-    {
-      "epoch": 0.15,
-      "learning_rate": 9.460000000000001e-06,
-      "loss": 3.5401,
-      "step": 750
-    },
-    {
-      "epoch": 0.15,
-      "learning_rate": 9.404444444444445e-06,
-      "loss": 3.7507,
-      "step": 775
-    },
-    {
-      "epoch": 1.0,
-      "learning_rate": 9.348888888888889e-06,
-      "loss": 3.3527,
-      "step": 800
-    },
-    {
-      "epoch": 1.01,
-      "learning_rate": 9.293333333333335e-06,
-      "loss": 3.4445,
-      "step": 825
-    },
-    {
-      "epoch": 1.01,
-      "learning_rate": 9.237777777777779e-06,
-      "loss": 3.3492,
-      "step": 850
-    },
-    {
-      "epoch": 1.02,
-      "learning_rate": 9.182222222222223e-06,
-      "loss": 3.4195,
-      "step": 875
-    },
-    {
-      "epoch": 1.02,
-      "learning_rate": 9.126666666666667e-06,
-      "loss": 3.6324,
-      "step": 900
-    },
-    {
-      "epoch": 1.03,
-      "learning_rate": 9.07111111111111e-06,
-      "loss": 3.5551,
-      "step": 925
-    },
-    {
-      "epoch": 1.03,
-      "learning_rate": 9.015555555555557e-06,
-      "loss": 3.6139,
-      "step": 950
-    },
-    {
-      "epoch": 1.04,
-      "learning_rate": 8.96e-06,
-      "loss": 3.3878,
-      "step": 975
-    },
-    {
-      "epoch": 1.04,
-      "learning_rate": 8.904444444444446e-06,
-      "loss": 3.5696,
       "step": 1000
     },
     {
-      "epoch": 1.04,
-      "eval_cer": 96.89103531542409,
-      "eval_loss": 3.418963670730591,
-      "eval_runtime": 273.6647,
-      "eval_samples_per_second": 3.603,
-      "eval_steps_per_second": 1.801,
-      "eval_wer": 96.85496183206106,
       "step": 1000
     },
     {
-      "epoch": 1.05,
-      "learning_rate": 8.848888888888889e-06,
-      "loss": 3.6154,
-      "step": 1025
-    },
-    {
-      "epoch": 1.05,
-      "learning_rate": 8.793333333333334e-06,
-      "loss": 3.4399,
-      "step": 1050
-    },
-    {
-      "epoch": 1.06,
-      "learning_rate": 8.737777777777778e-06,
-      "loss": 3.3262,
-      "step": 1075
-    },
-    {
-      "epoch": 1.06,
-      "learning_rate": 8.682222222222222e-06,
-      "loss": 3.4615,
-      "step": 1100
-    },
-    {
-      "epoch": 1.07,
-      "learning_rate": 8.626666666666668e-06,
-      "loss": 3.2971,
-      "step": 1125
-    },
-    {
-      "epoch": 1.07,
-      "learning_rate": 8.571111111111112e-06,
-      "loss": 3.4235,
-      "step": 1150
-    },
-    {
-      "epoch": 1.08,
-      "learning_rate": 8.515555555555556e-06,
-      "loss": 3.252,
-      "step": 1175
-    },
-    {
-      "epoch": 1.08,
-      "learning_rate": 8.46e-06,
-      "loss": 3.3725,
-      "step": 1200
-    },
-    {
-      "epoch": 1.09,
-      "learning_rate": 8.404444444444444e-06,
-      "loss": 3.4539,
-      "step": 1225
-    },
-    {
-      "epoch": 1.09,
-      "learning_rate": 8.34888888888889e-06,
-      "loss": 3.3736,
-      "step": 1250
-    },
-    {
-      "epoch": 1.1,
-      "learning_rate": 8.293333333333334e-06,
-      "loss": 3.3606,
-      "step": 1275
-    },
-    {
-      "epoch": 1.1,
-      "learning_rate": 8.237777777777778e-06,
-      "loss": 3.4193,
-      "step": 1300
-    },
-    {
-      "epoch": 1.11,
-      "learning_rate": 8.182222222222222e-06,
-      "loss": 3.3679,
-      "step": 1325
-    },
-    {
-      "epoch": 1.11,
-      "learning_rate": 8.126666666666668e-06,
-      "loss": 3.5329,
-      "step": 1350
-    },
-    {
-      "epoch": 1.12,
-      "learning_rate": 8.071111111111112e-06,
-      "loss": 3.3347,
-      "step": 1375
-    },
-    {
-      "epoch": 1.12,
-      "learning_rate": 8.015555555555556e-06,
-      "loss": 3.3093,
-      "step": 1400
-    },
-    {
-      "epoch": 1.13,
-      "learning_rate": 7.960000000000002e-06,
-      "loss": 3.3035,
-      "step": 1425
-    },
-    {
-      "epoch": 1.13,
-      "learning_rate": 7.904444444444444e-06,
-      "loss": 3.4493,
-      "step": 1450
-    },
-    {
-      "epoch": 1.14,
-      "learning_rate": 7.84888888888889e-06,
-      "loss": 3.2347,
-      "step": 1475
-    },
-    {
-      "epoch": 1.14,
-      "learning_rate": 7.793333333333334e-06,
-      "loss": 3.2076,
-      "step": 1500
-    },
-    {
-      "epoch": 1.15,
-      "learning_rate": 7.737777777777778e-06,
-      "loss": 3.2862,
-      "step": 1525
-    },
-    {
-      "epoch": 1.15,
-      "learning_rate": 7.682222222222224e-06,
-      "loss": 3.4486,
-      "step": 1550
-    },
-    {
-      "epoch": 1.16,
-      "learning_rate": 7.626666666666668e-06,
-      "loss": 3.3531,
-      "step": 1575
-    },
-    {
-      "epoch": 2.0,
-      "learning_rate": 7.571111111111112e-06,
-      "loss": 3.435,
-      "step": 1600
-    },
-    {
-      "epoch": 2.01,
-      "learning_rate": 7.515555555555556e-06,
-      "loss": 3.3428,
-      "step": 1625
-    },
-    {
-      "epoch": 2.01,
-      "learning_rate": 7.4600000000000006e-06,
-      "loss": 3.2712,
-      "step": 1650
-    },
-    {
-      "epoch": 2.02,
-      "learning_rate": 7.4044444444444455e-06,
-      "loss": 3.147,
-      "step": 1675
-    },
-    {
-      "epoch": 2.02,
-      "learning_rate": 7.3488888888888895e-06,
-      "loss": 3.2841,
-      "step": 1700
-    },
-    {
-      "epoch": 2.03,
-      "learning_rate": 7.2933333333333335e-06,
-      "loss": 3.1548,
-      "step": 1725
-    },
-    {
-      "epoch": 2.03,
-      "learning_rate": 7.237777777777778e-06,
-      "loss": 3.0778,
-      "step": 1750
-    },
-    {
-      "epoch": 2.04,
-      "learning_rate": 7.1822222222222224e-06,
-      "loss": 3.1293,
-      "step": 1775
-    },
-    {
-      "epoch": 2.04,
-      "learning_rate": 7.126666666666667e-06,
-      "loss": 3.3325,
-      "step": 1800
-    },
-    {
-      "epoch": 2.05,
-      "learning_rate": 7.071111111111112e-06,
-      "loss": 3.1438,
-      "step": 1825
-    },
-    {
-      "epoch": 2.05,
-      "learning_rate": 7.015555555555556e-06,
-      "loss": 3.3838,
-      "step": 1850
-    },
-    {
-      "epoch": 2.06,
-      "learning_rate": 6.96e-06,
-      "loss": 3.344,
-      "step": 1875
-    },
-    {
-      "epoch": 2.06,
-      "learning_rate": 6.904444444444444e-06,
-      "loss": 3.1949,
-      "step": 1900
-    },
-    {
-      "epoch": 2.07,
-      "learning_rate": 6.848888888888889e-06,
-      "loss": 3.2732,
-      "step": 1925
-    },
-    {
-      "epoch": 2.07,
-      "learning_rate": 6.793333333333334e-06,
-      "loss": 3.2508,
-      "step": 1950
-    },
-    {
-      "epoch": 2.08,
-      "learning_rate": 6.737777777777779e-06,
-      "loss": 3.1962,
-      "step": 1975
-    },
-    {
-      "epoch": 2.08,
-      "learning_rate": 6.682222222222223e-06,
-      "loss": 3.1453,
-      "step": 2000
-    },
-    {
-      "epoch": 2.08,
-      "eval_cer": 98.94355568970722,
-      "eval_loss": 3.238253116607666,
-      "eval_runtime": 276.3056,
-      "eval_samples_per_second": 3.569,
-      "eval_steps_per_second": 1.784,
-      "eval_wer": 98.93129770992365,
-      "step": 2000
-    },
-    {
-      "epoch": 2.09,
-      "learning_rate": 6.626666666666667e-06,
-      "loss": 3.2259,
-      "step": 2025
-    },
-    {
-      "epoch": 2.09,
-      "learning_rate": 6.571111111111111e-06,
-      "loss": 3.2412,
-      "step": 2050
-    },
-    {
-      "epoch": 2.1,
-      "learning_rate": 6.515555555555556e-06,
-      "loss": 3.1211,
-      "step": 2075
-    },
-    {
-      "epoch": 2.1,
-      "learning_rate": 6.460000000000001e-06,
-      "loss": 2.9139,
-      "step": 2100
-    },
-    {
-      "epoch": 2.11,
-      "learning_rate": 6.404444444444446e-06,
-      "loss": 3.1701,
-      "step": 2125
-    },
-    {
-      "epoch": 2.11,
-      "learning_rate": 6.348888888888889e-06,
-      "loss": 3.1944,
-      "step": 2150
-    },
-    {
-      "epoch": 2.12,
-      "learning_rate": 6.293333333333334e-06,
-      "loss": 3.1589,
-      "step": 2175
-    },
-    {
-      "epoch": 2.12,
-      "learning_rate": 6.237777777777778e-06,
-      "loss": 3.2229,
-      "step": 2200
-    },
-    {
-      "epoch": 2.13,
-      "learning_rate": 6.182222222222223e-06,
-      "loss": 3.2401,
-      "step": 2225
-    },
-    {
-      "epoch": 2.13,
-      "learning_rate": 6.126666666666668e-06,
-      "loss": 3.2551,
-      "step": 2250
-    },
-    {
-      "epoch": 2.14,
-      "learning_rate": 6.0711111111111125e-06,
-      "loss": 2.9634,
-      "step": 2275
-    },
-    {
-      "epoch": 2.14,
-      "learning_rate": 6.015555555555556e-06,
-      "loss": 3.0801,
-      "step": 2300
-    },
-    {
-      "epoch": 2.15,
-      "learning_rate": 5.9600000000000005e-06,
-      "loss": 3.1698,
-      "step": 2325
-    },
-    {
-      "epoch": 2.15,
-      "learning_rate": 5.9044444444444446e-06,
-      "loss": 3.1864,
-      "step": 2350
-    },
-    {
-      "epoch": 3.0,
-      "learning_rate": 5.8488888888888895e-06,
-      "loss": 3.1402,
-      "step": 2375
-    },
-    {
-      "epoch": 3.01,
-      "learning_rate": 5.793333333333334e-06,
-      "loss": 3.1443,
-      "step": 2400
-    },
-    {
-      "epoch": 3.01,
-      "learning_rate": 5.737777777777778e-06,
-      "loss": 2.9661,
-      "step": 2425
-    },
-    {
-      "epoch": 3.02,
-      "learning_rate": 5.682222222222222e-06,
-      "loss": 3.0652,
-      "step": 2450
-    },
-    {
-      "epoch": 3.02,
-      "learning_rate": 5.626666666666667e-06,
-      "loss": 3.0405,
-      "step": 2475
-    },
-    {
-      "epoch": 3.03,
-      "learning_rate": 5.571111111111111e-06,
-      "loss": 2.9499,
-      "step": 2500
-    },
-    {
-      "epoch": 3.03,
-      "learning_rate": 5.515555555555556e-06,
-      "loss": 3.1246,
-      "step": 2525
-    },
-    {
-      "epoch": 3.04,
-      "learning_rate": 5.460000000000001e-06,
-      "loss": 3.0122,
-      "step": 2550
-    },
-    {
-      "epoch": 3.04,
-      "learning_rate": 5.404444444444444e-06,
-      "loss": 3.0258,
-      "step": 2575
-    },
-    {
-      "epoch": 3.05,
-      "learning_rate": 5.348888888888889e-06,
-      "loss": 3.1414,
-      "step": 2600
-    },
-    {
-      "epoch": 3.05,
-      "learning_rate": 5.293333333333334e-06,
-      "loss": 3.1096,
-      "step": 2625
-    },
-    {
-      "epoch": 3.06,
-      "learning_rate": 5.237777777777778e-06,
-      "loss": 3.0937,
-      "step": 2650
-    },
-    {
-      "epoch": 3.06,
-      "learning_rate": 5.182222222222223e-06,
-      "loss": 3.0308,
-      "step": 2675
-    },
-    {
-      "epoch": 3.07,
-      "learning_rate": 5.126666666666668e-06,
-      "loss": 3.1702,
-      "step": 2700
-    },
-    {
-      "epoch": 3.07,
-      "learning_rate": 5.071111111111111e-06,
-      "loss": 3.1638,
-      "step": 2725
-    },
-    {
-      "epoch": 3.08,
-      "learning_rate": 5.015555555555556e-06,
-      "loss": 2.9025,
-      "step": 2750
-    },
-    {
-      "epoch": 3.08,
-      "learning_rate": 4.960000000000001e-06,
-      "loss": 3.0278,
-      "step": 2775
-    },
-    {
-      "epoch": 3.09,
-      "learning_rate": 4.904444444444445e-06,
-      "loss": 3.0397,
-      "step": 2800
-    },
-    {
-      "epoch": 3.09,
-      "learning_rate": 4.848888888888889e-06,
-      "loss": 3.0413,
-      "step": 2825
-    },
-    {
-      "epoch": 3.1,
-      "learning_rate": 4.793333333333334e-06,
-      "loss": 3.1721,
-      "step": 2850
-    },
-    {
-      "epoch": 3.1,
-      "learning_rate": 4.737777777777779e-06,
-      "loss": 2.9547,
-      "step": 2875
-    },
-    {
-      "epoch": 3.11,
-      "learning_rate": 4.682222222222223e-06,
-      "loss": 2.974,
-      "step": 2900
-    },
-    {
-      "epoch": 3.11,
-      "learning_rate": 4.626666666666667e-06,
-      "loss": 3.0297,
-      "step": 2925
-    },
-    {
-      "epoch": 3.12,
-      "learning_rate": 4.571111111111112e-06,
-      "loss": 2.9829,
-      "step": 2950
-    },
-    {
-      "epoch": 3.12,
-      "learning_rate": 4.515555555555556e-06,
-      "loss": 2.8869,
-      "step": 2975
-    },
-    {
-      "epoch": 3.13,
-      "learning_rate": 4.4600000000000005e-06,
-      "loss": 3.0722,
-      "step": 3000
-    },
-    {
-      "epoch": 3.13,
-      "eval_cer": 158.52701479022033,
-      "eval_loss": 3.20432710647583,
-      "eval_runtime": 318.685,
-      "eval_samples_per_second": 3.094,
-      "eval_steps_per_second": 1.547,
-      "eval_wer": 129.06870229007635,
-      "step": 3000
-    },
-    {
-      "epoch": 3.13,
-      "learning_rate": 4.404444444444445e-06,
-      "loss": 2.9035,
-      "step": 3025
-    },
-    {
-      "epoch": 3.14,
-      "learning_rate": 4.348888888888889e-06,
-      "loss": 2.9127,
-      "step": 3050
-    },
-    {
-      "epoch": 3.14,
-      "learning_rate": 4.2933333333333334e-06,
-      "loss": 3.1372,
-      "step": 3075
-    },
-    {
-      "epoch": 3.15,
-      "learning_rate": 4.2377777777777775e-06,
-      "loss": 2.9721,
-      "step": 3100
-    },
-    {
-      "epoch": 3.15,
-      "learning_rate": 4.182222222222222e-06,
-      "loss": 2.9846,
-      "step": 3125
-    },
-    {
-      "epoch": 3.16,
-      "learning_rate": 4.126666666666667e-06,
-      "loss": 3.0426,
-      "step": 3150
-    },
-    {
-      "epoch": 4.0,
-      "learning_rate": 4.071111111111111e-06,
-      "loss": 2.9058,
-      "step": 3175
-    },
-    {
-      "epoch": 4.01,
-      "learning_rate": 4.015555555555556e-06,
-      "loss": 2.8259,
-      "step": 3200
-    },
-    {
-      "epoch": 4.01,
-      "learning_rate": 3.96e-06,
-      "loss": 2.9893,
-      "step": 3225
-    },
-    {
-      "epoch": 4.02,
-      "learning_rate": 3.904444444444444e-06,
-      "loss": 2.9297,
-      "step": 3250
-    },
-    {
-      "epoch": 4.02,
-      "learning_rate": 3.848888888888889e-06,
-      "loss": 2.8989,
-      "step": 3275
-    },
-    {
-      "epoch": 4.03,
-      "learning_rate": 3.793333333333334e-06,
-      "loss": 2.9417,
-      "step": 3300
-    },
-    {
-      "epoch": 4.03,
-      "learning_rate": 3.737777777777778e-06,
-      "loss": 2.9383,
-      "step": 3325
-    },
-    {
-      "epoch": 4.04,
-      "learning_rate": 3.6822222222222225e-06,
-      "loss": 2.8867,
-      "step": 3350
-    },
-    {
-      "epoch": 4.04,
-      "learning_rate": 3.6266666666666674e-06,
-      "loss": 2.9945,
-      "step": 3375
-    },
-    {
-      "epoch": 4.05,
-      "learning_rate": 3.5711111111111114e-06,
-      "loss": 2.9077,
-      "step": 3400
-    },
-    {
-      "epoch": 4.05,
-      "learning_rate": 3.515555555555556e-06,
-      "loss": 2.8711,
-      "step": 3425
-    },
-    {
-      "epoch": 4.06,
-      "learning_rate": 3.46e-06,
-      "loss": 2.9767,
-      "step": 3450
-    },
-    {
-      "epoch": 4.06,
-      "learning_rate": 3.404444444444445e-06,
-      "loss": 2.8937,
-      "step": 3475
-    },
-    {
-      "epoch": 4.07,
-      "learning_rate": 3.3488888888888892e-06,
-      "loss": 2.9073,
-      "step": 3500
-    },
-    {
-      "epoch": 4.07,
-      "learning_rate": 3.2933333333333333e-06,
-      "loss": 2.7517,
-      "step": 3525
-    },
-    {
-      "epoch": 4.08,
-      "learning_rate": 3.237777777777778e-06,
-      "loss": 3.0988,
-      "step": 3550
-    },
-    {
-      "epoch": 4.08,
-      "learning_rate": 3.1822222222222226e-06,
-      "loss": 2.7826,
-      "step": 3575
-    },
-    {
-      "epoch": 4.09,
-      "learning_rate": 3.1266666666666667e-06,
-      "loss": 2.9641,
-      "step": 3600
-    },
-    {
-      "epoch": 4.09,
-      "learning_rate": 3.0711111111111115e-06,
-      "loss": 2.8948,
-      "step": 3625
-    },
-    {
-      "epoch": 4.1,
-      "learning_rate": 3.015555555555556e-06,
-      "loss": 2.8519,
-      "step": 3650
-    },
-    {
-      "epoch": 4.1,
-      "learning_rate": 2.96e-06,
-      "loss": 2.8796,
-      "step": 3675
-    },
-    {
-      "epoch": 4.11,
-      "learning_rate": 2.904444444444445e-06,
-      "loss": 2.9119,
-      "step": 3700
-    },
-    {
-      "epoch": 4.11,
-      "learning_rate": 2.8488888888888894e-06,
-      "loss": 2.8479,
-      "step": 3725
-    },
-    {
-      "epoch": 4.12,
-      "learning_rate": 2.7933333333333334e-06,
-      "loss": 2.9925,
-      "step": 3750
-    },
-    {
-      "epoch": 4.12,
-      "learning_rate": 2.7377777777777783e-06,
-      "loss": 2.9697,
-      "step": 3775
-    },
-    {
-      "epoch": 4.13,
-      "learning_rate": 2.6822222222222223e-06,
-      "loss": 2.9331,
-      "step": 3800
-    },
-    {
-      "epoch": 4.13,
-      "learning_rate": 2.6266666666666668e-06,
-      "loss": 2.7614,
-      "step": 3825
-    },
-    {
-      "epoch": 4.14,
-      "learning_rate": 2.5711111111111112e-06,
-      "loss": 2.9913,
-      "step": 3850
-    },
-    {
-      "epoch": 4.14,
-      "learning_rate": 2.5155555555555557e-06,
-      "loss": 2.8217,
-      "step": 3875
-    },
-    {
-      "epoch": 4.15,
-      "learning_rate": 2.46e-06,
-      "loss": 2.7742,
-      "step": 3900
-    },
-    {
-      "epoch": 4.15,
-      "learning_rate": 2.4044444444444446e-06,
-      "loss": 2.8652,
-      "step": 3925
-    },
-    {
-      "epoch": 5.0,
-      "learning_rate": 2.348888888888889e-06,
-      "loss": 2.9482,
-      "step": 3950
-    },
-    {
-      "epoch": 5.01,
-      "learning_rate": 2.2933333333333335e-06,
-      "loss": 2.7808,
-      "step": 3975
-    },
-    {
-      "epoch": 5.01,
-      "learning_rate": 2.237777777777778e-06,
-      "loss": 2.8327,
-      "step": 4000
-    },
-    {
-      "epoch": 5.01,
-      "eval_cer": 333.05161485058863,
-      "eval_loss": 3.225778102874756,
-      "eval_runtime": 362.3621,
-      "eval_samples_per_second": 2.721,
-      "eval_steps_per_second": 1.361,
-      "eval_wer": 327.9083969465649,
-      "step": 4000
-    },
-    {
-      "epoch": 5.02,
-      "learning_rate": 2.1822222222222225e-06,
-      "loss": 2.7936,
-      "step": 4025
-    },
-    {
-      "epoch": 5.02,
-      "learning_rate": 2.126666666666667e-06,
-      "loss": 2.7693,
-      "step": 4050
-    },
-    {
-      "epoch": 5.03,
-      "learning_rate": 2.0711111111111114e-06,
-      "loss": 2.7341,
-      "step": 4075
-    },
-    {
-      "epoch": 5.03,
-      "learning_rate": 2.0155555555555554e-06,
-      "loss": 2.7995,
-      "step": 4100
-    },
-    {
-      "epoch": 5.04,
-      "learning_rate": 1.9600000000000003e-06,
-      "loss": 2.78,
-      "step": 4125
-    },
-    {
-      "epoch": 5.04,
-      "learning_rate": 1.9044444444444445e-06,
-      "loss": 2.8048,
-      "step": 4150
-    },
-    {
-      "epoch": 5.05,
-      "learning_rate": 1.848888888888889e-06,
-      "loss": 2.6663,
-      "step": 4175
-    },
-    {
-      "epoch": 5.05,
-      "learning_rate": 1.7933333333333337e-06,
-      "loss": 2.9016,
-      "step": 4200
-    },
-    {
-      "epoch": 5.06,
-      "learning_rate": 1.737777777777778e-06,
-      "loss": 2.7117,
-      "step": 4225
-    },
-    {
-      "epoch": 5.06,
-      "learning_rate": 1.6822222222222224e-06,
-      "loss": 2.6716,
-      "step": 4250
-    },
-    {
-      "epoch": 5.07,
-      "learning_rate": 1.6266666666666666e-06,
-      "loss": 2.7193,
-      "step": 4275
-    },
-    {
-      "epoch": 5.07,
-      "learning_rate": 1.5711111111111113e-06,
-      "loss": 2.7704,
-      "step": 4300
-    },
-    {
-      "epoch": 5.08,
-      "learning_rate": 1.5155555555555558e-06,
-      "loss": 2.9998,
-      "step": 4325
-    },
-    {
-      "epoch": 5.08,
-      "learning_rate": 1.46e-06,
-      "loss": 2.8433,
-      "step": 4350
-    },
-    {
-      "epoch": 5.09,
-      "learning_rate": 1.4044444444444447e-06,
-      "loss": 2.6071,
-      "step": 4375
-    },
-    {
-      "epoch": 5.09,
-      "learning_rate": 1.3488888888888891e-06,
-      "loss": 2.8095,
-      "step": 4400
-    },
-    {
-      "epoch": 5.1,
-      "learning_rate": 1.2933333333333334e-06,
-      "loss": 2.6704,
-      "step": 4425
-    },
-    {
-      "epoch": 5.1,
-      "learning_rate": 1.2377777777777778e-06,
-      "loss": 2.6676,
-      "step": 4450
-    },
-    {
-      "epoch": 5.11,
-      "learning_rate": 1.1822222222222223e-06,
-      "loss": 2.8023,
-      "step": 4475
-    },
-    {
-      "epoch": 5.11,
-      "learning_rate": 1.1266666666666667e-06,
-      "loss": 2.7526,
-      "step": 4500
-    },
-    {
-      "epoch": 5.12,
-      "learning_rate": 1.0711111111111112e-06,
-      "loss": 2.7228,
-      "step": 4525
-    },
-    {
-      "epoch": 5.12,
-      "learning_rate": 1.0155555555555557e-06,
-      "loss": 2.6382,
-      "step": 4550
-    },
-    {
-      "epoch": 5.13,
-      "learning_rate": 9.600000000000001e-07,
-      "loss": 2.7358,
-      "step": 4575
-    },
-    {
-      "epoch": 5.13,
-      "learning_rate": 9.044444444444445e-07,
-      "loss": 2.6922,
-      "step": 4600
-    },
-    {
-      "epoch": 5.14,
-      "learning_rate": 8.488888888888889e-07,
-      "loss": 2.847,
-      "step": 4625
-    },
-    {
-      "epoch": 5.14,
-      "learning_rate": 7.933333333333335e-07,
-      "loss": 2.6146,
-      "step": 4650
-    },
-    {
-      "epoch": 5.15,
-      "learning_rate": 7.377777777777779e-07,
-      "loss": 2.7914,
-      "step": 4675
-    },
-    {
-      "epoch": 5.15,
-      "learning_rate": 6.822222222222223e-07,
-      "loss": 2.645,
-      "step": 4700
-    },
-    {
-      "epoch": 5.16,
-      "learning_rate": 6.266666666666667e-07,
-      "loss": 2.717,
-      "step": 4725
-    },
-    {
-      "epoch": 6.0,
-      "learning_rate": 5.711111111111111e-07,
-      "loss": 2.751,
-      "step": 4750
-    },
-    {
-      "epoch": 6.01,
-      "learning_rate": 5.155555555555556e-07,
-      "loss": 2.6859,
-      "step": 4775
-    },
-    {
-      "epoch": 6.01,
-      "learning_rate": 4.6000000000000004e-07,
-      "loss": 2.4752,
-      "step": 4800
-    },
-    {
-      "epoch": 6.02,
-      "learning_rate": 4.0444444444444445e-07,
-      "loss": 2.6502,
-      "step": 4825
-    },
-    {
-      "epoch": 6.02,
-      "learning_rate": 3.488888888888889e-07,
-      "loss": 2.5226,
-      "step": 4850
-    },
-    {
-      "epoch": 6.03,
-      "learning_rate": 2.9333333333333337e-07,
-      "loss": 2.6241,
-      "step": 4875
-    },
-    {
-      "epoch": 6.03,
-      "learning_rate": 2.3777777777777777e-07,
-      "loss": 2.5514,
-      "step": 4900
-    },
-    {
-      "epoch": 6.04,
-      "learning_rate": 1.8222222222222226e-07,
-      "loss": 2.6996,
-      "step": 4925
-    },
-    {
-      "epoch": 6.04,
-      "learning_rate": 1.2666666666666666e-07,
-      "loss": 2.6358,
-      "step": 4950
-    },
-    {
-      "epoch": 6.05,
-      "learning_rate": 7.111111111111112e-08,
-      "loss": 2.7597,
-      "step": 4975
-    },
-    {
-      "epoch": 6.05,
-      "learning_rate": 1.5555555555555557e-08,
-      "loss": 2.6468,
-      "step": 5000
-    },
-    {
-      "epoch": 6.05,
-      "eval_cer": 123.45306368849984,
-      "eval_loss": 3.212874174118042,
-      "eval_runtime": 302.4073,
-      "eval_samples_per_second": 3.261,
-      "eval_steps_per_second": 1.63,
-      "eval_wer": 118.50381679389312,
-      "step": 5000
-    },
-    {
-      "epoch": 6.05,
-      "step": 5000,
-      "total_flos": 2.12189711818752e+19,
-      "train_loss": 3.0522214595794677,
-      "train_runtime": 5458.8622,
-      "train_samples_per_second": 1.832,
-      "train_steps_per_second": 0.916
     }
   ],
-  "max_steps": 5000,
   "num_train_epochs": 9223372036854775807,
-  "total_flos": 2.12189711818752e+19,
   "trial_name": null,
   "trial_params": null
 }

 {
   "best_metric": null,
   "best_model_checkpoint": null,
+  "epoch": 1.383,
+  "global_step": 1000,
   "is_hyper_param_search": false,
   "is_local_process_zero": true,
   "is_world_process_zero": true,
   "log_history": [
     {
+      "epoch": 1.38,
+      "learning_rate": 8.88888888888889e-08,
+      "loss": 0.9198,
       "step": 1000
     },
     {
+      "epoch": 1.38,
+      "eval_cer": 48.17386054935105,
+      "eval_loss": 0.9147109389305115,
+      "eval_runtime": 340.3264,
+      "eval_samples_per_second": 2.897,
+      "eval_steps_per_second": 1.449,
+      "eval_wer": 48.06106870229008,
       "step": 1000
     },
     {
+      "epoch": 1.38,
+      "step": 1000,
+      "total_flos": 4.2463420416e+18,
+      "train_loss": 0.91982373046875,
+      "train_runtime": 1347.6358,
+      "train_samples_per_second": 1.484,
+      "train_steps_per_second": 0.742
     }
   ],
+  "max_steps": 1000,
   "num_train_epochs": 9223372036854775807,
+  "total_flos": 4.2463420416e+18,
   "trial_name": null,
   "trial_params": null
 }

training_args.bin CHANGED Viewed

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:b0917c506d33e6aaecd26c5c143567a69a21ec072808d7d6a3e64a1189180b2d
-size 3515

 version https://git-lfs.github.com/spec/v1
+oid sha256:559c4ce05e89d747196b9bba18aa9426caf585ced1042f6a3e013da85ed0c8dc
+size 3707

vocab.json CHANGED Viewed

The diff for this file is too large to render. See raw diff