Skip to content

fix(distill): engram-resilient resume #221

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Merged
ronaldtse merged 1 commit into from
Sep 28, 2026
Merged

fix(distill): engram-resilient resume #221

Changes from all commits
Commits
Show all changes
1 commit
Select commit
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
15 changes: 12 additions & 3 deletions src/gpu/modal_distill.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -1149,9 +1149,18 @@ def _usable(ck: Path) -> bool:
key=lambda p: int(p.name.split("-")[1]),
)
if ckpts:
student.load_state_dict(
torch.load(ckpts[-1] / "student.pt", map_location="cpu", weights_only=True)
)
sd = torch.load(ckpts[-1] / "student.pt", map_location="cpu", weights_only=True)
missing, unexpected = student.load_state_dict(sd, strict=False)
# engram params may be absent in older checkpoints: the module
# is zero-init (identity), so a fresh table is a safe resume.
# Anything else missing or unexpected is a real mismatch.
bad_missing = [k for k in missing if not k.startswith("_engram.")]
if bad_missing or unexpected:
raise RuntimeError(
f"resume mismatch: missing={bad_missing} unexpected={list(unexpected)}"
)
if missing:
print(f"[{spec_id}] resume with fresh engram: {missing}", flush=True)
optimizer.load_state_dict(
torch.load(ckpts[-1] / "optim.pt", map_location="cpu", weights_only=True)
)
Expand Down
Loading