Instructions to use tobiges/behavior_fast with libraries, inference providers, notebooks, and local apps. Follow these links to get started.
- Libraries
- Transformers
How to use tobiges/behavior_fast with Transformers:
# Load model directly from transformers import AutoModel model = AutoModel.from_pretrained("tobiges/behavior_fast", device_map="auto") - Notebooks
- Google Colab
- Kaggle
Upload processor
Browse files- processing_action_tokenizer.py +2 -2
- tokenizer.json +0 -0
processing_action_tokenizer.py
CHANGED
|
@@ -145,7 +145,7 @@ class UniversalActionProcessor(ProcessorMixin):
|
|
| 145 |
|
| 146 |
# Train BPE tokenizer
|
| 147 |
tokenizer = Tokenizer(BPE())
|
| 148 |
-
tokenizer.pre_tokenizer = pre_tokenizers.ByteLevel(add_prefix_space=False)
|
| 149 |
tokenizer.decoder = decoders.ByteLevel()
|
| 150 |
tokenizer.post_processor = processors.ByteLevel(trim_offsets=False)
|
| 151 |
|
|
@@ -157,7 +157,7 @@ class UniversalActionProcessor(ProcessorMixin):
|
|
| 157 |
show_progress=True,
|
| 158 |
special_tokens=[],
|
| 159 |
initial_alphabet=pre_tokenizers.ByteLevel.alphabet(),
|
| 160 |
-
max_token_length=
|
| 161 |
)
|
| 162 |
tokenizer.train_from_iterator(
|
| 163 |
_token_iter(), trainer=trainer, length=len(dct_tokens)
|
|
|
|
| 145 |
|
| 146 |
# Train BPE tokenizer
|
| 147 |
tokenizer = Tokenizer(BPE())
|
| 148 |
+
tokenizer.pre_tokenizer = pre_tokenizers.ByteLevel(add_prefix_space=False, use_regex=False)
|
| 149 |
tokenizer.decoder = decoders.ByteLevel()
|
| 150 |
tokenizer.post_processor = processors.ByteLevel(trim_offsets=False)
|
| 151 |
|
|
|
|
| 157 |
show_progress=True,
|
| 158 |
special_tokens=[],
|
| 159 |
initial_alphabet=pre_tokenizers.ByteLevel.alphabet(),
|
| 160 |
+
max_token_length=128,
|
| 161 |
)
|
| 162 |
tokenizer.train_from_iterator(
|
| 163 |
_token_iter(), trainer=trainer, length=len(dct_tokens)
|
tokenizer.json
CHANGED
|
The diff for this file is too large to render.
See raw diff
|
|
|