This commit is contained in:
Felix Marty 2024-06-21 16:11:38 +00:00 committed by Nicolas Patry
parent bc2b9b20e2
commit 1bb1a344d7
No known key found for this signature in database
GPG Key ID: E939E8CC91A1C674

View File

@ -40,15 +40,14 @@ def cleanup_cache(token: str):
size_per_model = {} size_per_model = {}
extension_per_model = {} extension_per_model = {}
for model_id, revision in REQUIRED_MODELS.items(): for model_id, revision in REQUIRED_MODELS.items():
print(f"Crawling {model_id}...")
model_size = 0 model_size = 0
print(f"call huggingface_hub.list_repo_files for {model_id}")
all_files = huggingface_hub.list_repo_files( all_files = huggingface_hub.list_repo_files(
model_id, model_id,
repo_type="model", repo_type="model",
revision=revision, revision=revision,
token=token, token=token,
) )
print(f"end huggingface_hub.list_repo_files")
extension = None extension = None
if any(".safetensors" in filename for filename in all_files): if any(".safetensors" in filename for filename in all_files):
@ -60,9 +59,6 @@ def cleanup_cache(token: str):
extension_per_model[model_id] = extension extension_per_model[model_id] = extension
print(
f"call huggingface_hub.hf_hub_url & huggingface_hub.get_hf_file_metadata for {model_id}"
)
for filename in all_files: for filename in all_files:
if filename.endswith(extension): if filename.endswith(extension):
file_url = huggingface_hub.hf_hub_url( file_url = huggingface_hub.hf_hub_url(
@ -72,7 +68,6 @@ def cleanup_cache(token: str):
file_url, token=token file_url, token=token
) )
model_size += file_metadata.size * 1e-9 # in GB model_size += file_metadata.size * 1e-9 # in GB
print(f"end two calls")
size_per_model[model_id] = model_size size_per_model[model_id] = model_size