Skip to content

Fix PI0 embed_tokens.weight loading + WandB list metric logging #3116

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

Open
he-yufeng wants to merge 1 commit into
base: main
Choose a base branch
from
Open

Fix PI0 embed_tokens.weight loading + WandB list metric logging #3116

Show file tree
Hide file tree
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
10 changes: 10 additions & 0 deletions src/lerobot/policies/pi0/modeling_pi0.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -1130,10 +1130,20 @@ def _fix_pytorch_state_dict_keys(
key == "model.paligemma_with_expert.paligemma.lm_head.weight"
or key == "paligemma_with_expert.paligemma.lm_head.weight"
):
# PaLI-Gemma ties lm_head and embed_tokens weights
fixed_state_dict[
"model.paligemma_with_expert.paligemma.model.language_model.embed_tokens.weight"
] = value.clone()

# Also handle checkpoints that store embed_tokens.weight directly
if key in (
"paligemma_with_expert.paligemma.model.language_model.embed_tokens.weight",
"model.paligemma_with_expert.paligemma.model.language_model.embed_tokens.weight",
):
fixed_state_dict[
"model.paligemma_with_expert.paligemma.model.language_model.embed_tokens.weight"
] = value

fixed_state_dict[new_key] = value

return fixed_state_dict
Expand Down
7 changes: 7 additions & 0 deletions src/lerobot/rl/wandb_utils.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -177,6 +177,13 @@ def log_dict(
self._wandb.define_metric(new_custom_key, hidden=True)

for k, v in d.items():
# Log list metrics as individual entries (e.g. loss_per_dim)
if isinstance(v, (list, tuple)):
for i, elem in enumerate(v):
if isinstance(elem, (int, float)):
d_expanded = {f"{mode}/{k}_{i}": elem}
self._wandb.log(d_expanded)
continue
if not isinstance(v, (int | float | str)):
logging.warning(
f'WandB logging of key "{k}" was ignored as its type "{type(v)}" is not handled by this wrapper.'
Expand Down