Skip to content

fix: gemma-3 checkpoint conversion from litgpt to hf #2195

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
lianakoleva merged 4 commits into from
Jan 23, 2026
Merged

fix: gemma-3 checkpoint conversion from litgpt to hf #2195

Changes from 2 commits
Commits
Show all changes
4 commits
Select commit Hold shift + click to select a range
6b83332
Fix: Correct Gemma-3 checkpoint conversion
adi776borate Jan 18, 2026
47bd7e7
Merge branch 'main' into fix/gemma3-convert-lit
bhimrazy Jan 18, 2026
b923101
Fix Gemma weight conversion defaults and reorder conditions
adi776borate Jan 20, 2026
33b126d
Merge branch 'fix/gemma3-convert-lit' of https://github.com/adi776bor…
adi776borate Jan 20, 2026
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
3 changes: 3 additions & 0 deletions litgpt/scripts/convert_lit_checkpoint.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -565,6 +565,9 @@ def convert_lit_checkpoint(checkpoint_dir: Path, output_dir: Path) -> None:
copy_fn = partial(copy_weights_olmo2, config)
elif config.name.lower().startswith("qwen3"):
copy_fn = partial(copy_weights_qwen_3, config)
elif config.name.startswith("Gemma-3"):
untie_weights = True
copy_fn = partial(copy_weights_gemma_3, config, untie_weights=untie_weights)
elif config.mlp_class_name in ("LLaMAMLP", "GemmaMLP", "LLaMAMoE"):
untie_weights = "Gemma" in config.name
copy_fn = partial(copy_weights_llama, config, untie_weights=untie_weights)
Expand Down