Remove deprecated tokenizer config fields
Browse files
preprocessor/tokenizer_config.yaml
CHANGED
|
@@ -1,17 +1,17 @@
|
|
| 1 |
name: whisper_bpe_tokenizer
|
| 2 |
config_type: preprocessor
|
| 3 |
-
max_length: 448
|
| 4 |
-
truncation: longest_first
|
| 5 |
truncation_side: right
|
| 6 |
-
stride: 0
|
| 7 |
-
padding: longest
|
| 8 |
padding_side: right
|
|
|
|
| 9 |
pad_to_multiple_of: 0
|
| 10 |
pad_token_type_id: 0
|
| 11 |
bos_token: <|startoftranscript|>
|
| 12 |
eos_token: <|endoftext|>
|
| 13 |
unk_token: <|endoftext|>
|
|
|
|
| 14 |
pad_token: <|endoftext|>
|
|
|
|
|
|
|
| 15 |
additional_special_tokens:
|
| 16 |
- <|endoftext|>
|
| 17 |
- <|endoftext|>
|
|
|
|
| 1 |
name: whisper_bpe_tokenizer
|
| 2 |
config_type: preprocessor
|
|
|
|
|
|
|
| 3 |
truncation_side: right
|
|
|
|
|
|
|
| 4 |
padding_side: right
|
| 5 |
+
stride: 0
|
| 6 |
pad_to_multiple_of: 0
|
| 7 |
pad_token_type_id: 0
|
| 8 |
bos_token: <|startoftranscript|>
|
| 9 |
eos_token: <|endoftext|>
|
| 10 |
unk_token: <|endoftext|>
|
| 11 |
+
sep_token: <sep>
|
| 12 |
pad_token: <|endoftext|>
|
| 13 |
+
cls_token: <cls>
|
| 14 |
+
mask_token: <mask>
|
| 15 |
additional_special_tokens:
|
| 16 |
- <|endoftext|>
|
| 17 |
- <|endoftext|>
|