Spaces:
Running on Zero
Running on Zero
File size: 1,997 Bytes
86e797b | 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 | """Cache pinned official adapter/base snapshots before loading model code."""
import json
from pathlib import Path
from huggingface_hub import HfApi, snapshot_download
from visual_doc_rag.data import write_json
MODEL_ID = "ModernVBERT/colmodernvbert"
def cache_model(data_dir: Path, revision: str = "main", base_revision: str | None = None) -> dict:
root = data_dir / "models"
manifest_path = root / "manifest.json"
if manifest_path.exists():
cached = json.loads(manifest_path.read_text(encoding="utf-8"))
if revision in {"main", cached["adapter_revision"]} and base_revision in {
None,
cached["base_revision"],
}:
return cached
api = HfApi()
adapter_sha = api.model_info(MODEL_ID, revision=revision).sha
adapter_dir = root / "adapter" / adapter_sha
snapshot_download(
MODEL_ID,
revision=adapter_sha,
local_dir=adapter_dir,
cache_dir=root / "hub",
allow_patterns=["*.json", "*.safetensors", "*.jinja", "README.md", "LICENSE*"],
max_workers=2,
)
adapter = json.loads((adapter_dir / "adapter_config.json").read_text(encoding="utf-8"))
base_id = adapter["base_model_name_or_path"]
base_sha = api.model_info(
base_id, revision=base_revision or adapter.get("revision") or "main"
).sha
base_dir = root / "base" / base_sha
snapshot_download(
base_id,
revision=base_sha,
local_dir=base_dir,
cache_dir=root / "hub",
allow_patterns=["*.json", "*.safetensors", "*.jinja", "README.md", "LICENSE*"],
max_workers=2,
)
manifest = {
"model_id": MODEL_ID,
"adapter_revision": adapter_sha,
"base_model_id": base_id,
"base_revision": base_sha,
"adapter_path": str(adapter_dir.relative_to(root)),
"base_path": str(base_dir.relative_to(root)),
"license": "MIT",
}
write_json(manifest_path, manifest)
return manifest
|