HF endpoint auto-fallback: when huggingface.co is unreachable, probe mirrors (hf-mirror.com) and switch automatically with a one-line notice; error only if all fail

Co-Authored-By: Claude <noreply@anthropic.com>
This commit is contained in:
sora 2026-09-10 10:57:16 +00:00
parent 8ea83fea4f
commit dd7e1933ff

View File

@ -462,22 +462,36 @@ def _load_from_hub(spec: DatasetSpec) -> List[Dict[str, Any]]:
return [dict(r) for r in ds]
_HF_MIRROR_FALLBACKS = ['https://hf-mirror.com']
def _probe_hub_reachable(spec: DatasetSpec) -> None:
"""Fail fast when the HF endpoint is unreachable -- otherwise the hub
client retries silently for minutes and looks like a hang."""
client retries silently for minutes and looks like a hang.
When the DEFAULT endpoint is down but a mirror answers, switch to it
automatically (one-line notice, same behavior as --hf-endpoint)."""
base = _hf_base()
err = None
for candidate in [base] + (_HF_MIRROR_FALLBACKS if base == 'https://huggingface.co' else []):
try:
req = urllib.request.Request(base, method='HEAD',
req = urllib.request.Request(candidate, method='HEAD',
headers={'User-Agent': 'evalharness/0.1'})
urllib.request.urlopen(req, timeout=8)
if candidate != base:
os.environ['HF_ENDPOINT'] = candidate
print(f'· {base} unreachable -- falling back to {candidate}',
flush=True)
return
except Exception as e:
err = e
raise RuntimeError(
f'HuggingFace endpoint unreachable: {base}\n'
f' reason: {type(e).__name__}: {str(e)[:80]}\n'
f'HuggingFace endpoint unreachable: {base}'
+ (f' (mirrors tried: {", ".join(_HF_MIRROR_FALLBACKS)})'
if base == 'https://huggingface.co' else '')
+ f'\n reason: {type(err).__name__}: {str(err)[:80]}\n'
f' dataset {spec.source!r} cannot download. Fix:\n'
f' evalharness eval run ... --hf-endpoint https://hf-mirror.com\n'
f' (or export HF_ENDPOINT=https://hf-mirror.com)') from e
f' evalharness eval run ... --hf-endpoint <a reachable endpoint>') from err
def field_spec_to_record_fn(fields: FieldSpec) -> Callable[[Dict[str, Any]], Sample]: