Add a tokenizer placeholder

This commit is contained in:
oobabooga 2023-03-03 15:16:55 -03:00
parent 4cc36dc434
commit a345a2acd2

View File

@ -21,8 +21,11 @@ def get_max_prompt_length(tokens):
return max_length
def encode(prompt, tokens_to_generate=0, add_special_tokens=True):
# These models do not have explicit tokenizers for now, so
# we return an estimate on the number of tokens
if shared.is_RWKV or shared.is_LLaMA:
return prompt
return np.zeros((1, len(prompt)//5))
input_ids = shared.tokenizer.encode(str(prompt), return_tensors='pt', truncation=True, max_length=get_max_prompt_length(tokens_to_generate), add_special_tokens=add_special_tokens)
if shared.args.cpu: