download tokenizer when present

This commit is contained in:
Ber Zoidberg 2023-03-09 19:08:09 -08:00
parent 7c3d1b43c1
commit ec3de0495c

View File

@ -107,9 +107,10 @@ def get_download_links_from_huggingface(model, branch):
is_pytorch = re.match("pytorch_model.*\.bin", fname)
is_safetensors = re.match("model.*\.safetensors", fname)
is_tokenizer = re.match("tokenizer.*\.model", fname)
is_text = re.match(".*\.(txt|json)", fname)
if is_text or is_safetensors or is_pytorch:
if any((is_pytorch, is_safetensors, is_text, is_tokenizer)):
if is_text:
links.append(f"https://huggingface.co/{model}/resolve/{branch}/{fname}")
classifications.append('text')