Fix loading condition for universal llama tokenizer (#2753)

This commit is contained in:
ThisIsPIRI 2023-06-18 21:14:06 +00:00 committed by GitHub
parent 490a1795f0
commit def3b69002
No known key found for this signature in database
GPG Key ID: 4AEE18F83AFDEB23

View File

@ -86,7 +86,7 @@ def load_tokenizer(model_name, model):
tokenizer = AutoTokenizer.from_pretrained(Path(f"{shared.args.model_dir}/gpt-j-6B/"))
elif type(model) is transformers.LlamaForCausalLM or "LlamaGPTQForCausalLM" in str(type(model)):
# Try to load an universal LLaMA tokenizer
if any(s in shared.model_name.lower() for s in ['llava', 'oasst']):
if not any(s in shared.model_name.lower() for s in ['llava', 'oasst']):
for p in [Path(f"{shared.args.model_dir}/llama-tokenizer/"), Path(f"{shared.args.model_dir}/oobabooga_llama-tokenizer/")]:
if p.exists():
logger.info(f"Loading the universal LLaMA tokenizer from {p}...")