Update tokenization_phi3_small.py (#18)
Browse files- Update tokenization_phi3_small.py (3558f70d46aadc019146e8744232c2374234b8c1)
Co-authored-by: David Majercak <[email protected]>
tokenization_phi3_small.py
CHANGED
|
@@ -180,12 +180,10 @@ class Phi3SmallTokenizer(PreTrainedTokenizer):
|
|
| 180 |
# First try to load from the tokenization config if it exists
|
| 181 |
tokenization_config = get_tokenizer_config(pretrained_model_name_or_path, **kwargs)
|
| 182 |
if tokenization_config:
|
| 183 |
-
cls_kwargs
|
| 184 |
-
|
| 185 |
-
|
| 186 |
-
|
| 187 |
-
)
|
| 188 |
-
)
|
| 189 |
else:
|
| 190 |
config = AutoConfig.from_pretrained(pretrained_model_name_or_path, trust_remote_code=True)
|
| 191 |
cls_kwargs["model_max_length"] = config.max_position_embeddings
|
|
|
|
| 180 |
# First try to load from the tokenization config if it exists
|
| 181 |
tokenization_config = get_tokenizer_config(pretrained_model_name_or_path, **kwargs)
|
| 182 |
if tokenization_config:
|
| 183 |
+
cls_kwargs = {
|
| 184 |
+
**tokenization_config,
|
| 185 |
+
**cls_kwargs
|
| 186 |
+
}
|
|
|
|
|
|
|
| 187 |
else:
|
| 188 |
config = AutoConfig.from_pretrained(pretrained_model_name_or_path, trust_remote_code=True)
|
| 189 |
cls_kwargs["model_max_length"] = config.max_position_embeddings
|