Spaces:

eson
/

tokenizer-arena

Running

fix unicode error: 'unicodeescape' codec can't decode bytes in position 602-608: unknown Unicode character name

bce41d0 4 months ago

No virus

384 Bytes

	"""
	## 默认 use_fast=True 报错
	lib\site-packages\transformers\tokenization_utils_fast.py", line 504, in _batch_encode_plus
	encodings = self._tokenizer.encode_batch(
	pyo3_runtime.PanicException: AddedVocabulary bad split
	"""
	from transformers import AutoTokenizer

	tokenizer = AutoTokenizer.from_pretrained("lmsys/fastchat-t5-3b-v1.0", trust_remote_code=True, use_fast=False)