Delete scripts/old-download_ckpts.py
Browse files
scripts/old-download_ckpts.py
DELETED
|
@@ -1,82 +0,0 @@
|
|
| 1 |
-
import os
|
| 2 |
-
import argparse
|
| 3 |
-
import yaml
|
| 4 |
-
from huggingface_hub import snapshot_download
|
| 5 |
-
|
| 6 |
-
def download_repository(repo_id, local_dir, token=None):
|
| 7 |
-
"""
|
| 8 |
-
Download a Hugging Face repository to a local directory using snapshot_download.
|
| 9 |
-
|
| 10 |
-
Args:
|
| 11 |
-
repo_id (str): Hugging Face repository ID.
|
| 12 |
-
local_dir (str): Local directory to save the downloaded files.
|
| 13 |
-
token (str, optional): Hugging Face API token for private/gated repositories.
|
| 14 |
-
"""
|
| 15 |
-
os.makedirs(local_dir, exist_ok=True)
|
| 16 |
-
|
| 17 |
-
try:
|
| 18 |
-
snapshot_download(
|
| 19 |
-
repo_id=repo_id,
|
| 20 |
-
local_dir=local_dir,
|
| 21 |
-
local_dir_use_symlinks=False,
|
| 22 |
-
repo_type="model",
|
| 23 |
-
token=token if token else None,
|
| 24 |
-
allow_patterns=["*.safetensors", "*.ckpt", "*.json", "*.txt"]
|
| 25 |
-
)
|
| 26 |
-
print(f"Successfully downloaded all files from {repo_id} to {local_dir}")
|
| 27 |
-
except Exception as e:
|
| 28 |
-
print(f"Error downloading repository {repo_id}: {str(e)}")
|
| 29 |
-
|
| 30 |
-
if __name__ == "__main__":
|
| 31 |
-
parser = argparse.ArgumentParser(description="Download Hugging Face repositories to local directories.")
|
| 32 |
-
parser.add_argument(
|
| 33 |
-
"--config",
|
| 34 |
-
type=str,
|
| 35 |
-
default="configs/model_ckpts.yaml",
|
| 36 |
-
help="Path to the YAML configuration file containing model IDs and local directories"
|
| 37 |
-
)
|
| 38 |
-
parser.add_argument(
|
| 39 |
-
"--token",
|
| 40 |
-
type=str,
|
| 41 |
-
default=None,
|
| 42 |
-
help="Hugging Face API token for private or gated repositories (optional)"
|
| 43 |
-
)
|
| 44 |
-
|
| 45 |
-
args = parser.parse_args()
|
| 46 |
-
|
| 47 |
-
# Load the YAML configuration file
|
| 48 |
-
try:
|
| 49 |
-
with open(args.config, "r") as f:
|
| 50 |
-
config_data = yaml.safe_load(f)
|
| 51 |
-
|
| 52 |
-
if not config_data:
|
| 53 |
-
raise ValueError("The YAML file is empty or invalid")
|
| 54 |
-
|
| 55 |
-
# Extract repo_ids and local_dirs for HuggingFace platform only
|
| 56 |
-
repo_ids = []
|
| 57 |
-
local_dirs = []
|
| 58 |
-
for item in config_data:
|
| 59 |
-
if item.get("platform") == "HuggingFace":
|
| 60 |
-
repo_ids.append(item["model_id"])
|
| 61 |
-
local_dirs.append(item["local_dir"])
|
| 62 |
-
|
| 63 |
-
# Validate that at least one valid entry was found
|
| 64 |
-
if not repo_ids or not local_dirs:
|
| 65 |
-
raise ValueError("No valid HuggingFace platform entries found in the YAML file")
|
| 66 |
-
|
| 67 |
-
# Validate that the number of repo_ids matches the number of local_dirs
|
| 68 |
-
if len(repo_ids) != len(local_dirs):
|
| 69 |
-
raise ValueError("The number of model_ids must match the number of local_dirs for HuggingFace platform")
|
| 70 |
-
|
| 71 |
-
# Download each repository
|
| 72 |
-
for repo_id, local_dir in zip(repo_ids, local_dirs):
|
| 73 |
-
download_repository(repo_id, local_dir, args.token)
|
| 74 |
-
|
| 75 |
-
except FileNotFoundError:
|
| 76 |
-
print(f"Error: The configuration file '{args.config}' was not found.")
|
| 77 |
-
except yaml.YAMLError as e:
|
| 78 |
-
print(f"Error: Failed to parse YAML file. Details: {e}")
|
| 79 |
-
except KeyError as e:
|
| 80 |
-
print(f"Error: Missing expected key in YAML data. Details: {e}")
|
| 81 |
-
except ValueError as e:
|
| 82 |
-
print(f"Error: {e}")
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|