glaiveai

/

glaive-function-calling-v1

like 64

If someone one day has the same issue, just use "trust_remote_code=True" as an argument of AutoTokenizer.from_pretrained

I kind of have a similar issue. I am getting this error. I have the argument 'trust_remote_code' set to True

 ~/.cache/huggingface/modules/transformers_modules/glaiveai/glaive-function-calling- 
v1/d94dd60c9ceaff581cf7e6b5f9982b4b1716ae16/replit_lm_tokenizer.py in vocab_size(self)
 71     @property
 72     def vocab_size(self):
---> 73         return self.sp_model.get_piece_size()
 74 
 75     def get_vocab(self):

AttributeError: 'ReplitLMTokenizer' object has no attribute 'sp_model'

I kind of have a similar issue. I am getting this error. I have the argument 'trust_remote_code' set to True

 ~/.cache/huggingface/modules/transformers_modules/glaiveai/glaive-function-calling- 
v1/d94dd60c9ceaff581cf7e6b5f9982b4b1716ae16/replit_lm_tokenizer.py in vocab_size(self)
 71     @property
 72     def vocab_size(self):
---> 73         return self.sp_model.get_piece_size()
 74 
 75     def get_vocab(self):

AttributeError: 'ReplitLMTokenizer' object has no attribute 'sp_model'

Same issue on colab

Mine has "trust remote code " set to true but its still having the error

Edit: my mistake i was passing the argument to Vllm, ive modified the init method to reflect that, will try again and let you know

def __init__(
    self,
    model: str,
    tokenizer: Optional[str] = None,
    tokenizer_mode: str = "auto",
    trust_remote_code: bool = True, # modified to run replit
    tensor_parallel_size: int = 1,
    dtype: str = "auto",
    quantization: Optional[str] = None,
    revision: Optional[str] = None,
    tokenizer_revision: Optional[str] = None,
    seed: int = 0,
    gpu_memory_utilization: float = 0.9,
    swap_space: int = 4,
    **kwargs,
) -> None:

Still same issue here for me

When the error happens,

it looks like self.sp_model is not yet set, in replit tokenizer. I got a bit further by swapping it around so that the super().__init line comes after

Original:
def init(self, vocab_file, bos_token=None, eos_token='<|endoftext|>', unk_token='<|unk|>', pad_token='<|pad|>', sep_token=None, sp_model_kwargs: Optional[Dict[str, Any]]=None, **kwargs) -> None:
self.sp_model_kwargs = {} if sp_model_kwargs is None else sp_model_kwargs
super().init(bos_token=bos_token, eos_token=eos_token, unk_token=unk_token, pad_token=pad_token, sep_token=sep_token, sp_model_kwargs=self.sp_model_kwargs, **kwargs)
self.vocab_file = vocab_file
self.sp_model = spm.SentencePieceProcessor(**self.sp_model_kwargs)
self.sp_model.Load(vocab_file)

Edited version:
def init(self, vocab_file, bos_token=None, eos_token='<|endoftext|>', unk_token='<|unk|>', pad_token='<|pad|>', sep_token=None, sp_model_kwargs: Optional[Dict[str, Any]]=None, **kwargs) -> None:
self.sp_model_kwargs = {} if sp_model_kwargs is None else sp_model_kwargs
self.vocab_file = vocab_file
self.sp_model = spm.SentencePieceProcessor(**self.sp_model_kwargs)
self.sp_model.Load(vocab_file)
super().init(bos_token=bos_token, eos_token=eos_token, unk_token=unk_token, pad_token=pad_token, sep_token=sep_token, sp_model_kwargs=self.sp_model_kwargs, **kwargs)

The next issue i faced was with the HF_prefixlm_converter

15 and 16
I needed to comment out line 15,16,23 and 24
#from transformers.models.bloom.modeling_bloom import _expand_mask as _expand_mask_bloom
#from transformers.models.bloom.modeling_bloom import _make_causal_mask as _make_causal_mask_bloom

23 and 24
#from transformers.models.opt.modeling_opt import _expand_mask as _expand_mask_opt
#from transformers.models.opt.modeling_opt import _make_causal_mask as _make_causal_mask_opt