Skip to content

Pull requests: google-deepmind/gemma

New pull request New
183 Open 356 Closed
183 Open 356 Closed
Author
Filter by author
Loading
Label
Filter by label
Loading
Use alt + click/return to exclude labels
or ⇧ + click/return for logical OR
Projects
Filter by project
Loading
Milestones
Filter by milestone
Loading
Reviews
Filter by reviews
No reviews Review required Approved review Changes requested
Assignee
Filter by who’s assigned
Assigned to nobody Loading
Sort
Sort by
Newest Oldest Most commented Least commented Recently updated Least recently updated Best match
Most reactions
👍 👎 😄 🎉 😕 ❤️ 🚀 👀

Pull requests list

Fix Gemma 3/3n KV cache to honor per-batch end_index
#765 opened Jul 28, 2026 by Ayush7614 Loading…
4 tasks done
1
Implement peft.fuse_params / unfuse_params for LoRA weight fusion
#764 opened Jul 28, 2026 by Ayush7614 Loading…
2 tasks done
1
Fix assorted docstring/comment typos across the codebase
#763 opened Jul 25, 2026 by PratikDhanave Loading…
Fix _maybe_unbatch return annotation (Float -> Array) in t5gemma
#762 opened Jul 25, 2026 by PratikDhanave Loading…
Fix comment referencing non-existent variable in sft_model
#761 opened Jul 25, 2026 by PratikDhanave Loading…
Fix QueryPreAttentionNormalisation docstring to describe query scaling
#760 opened Jul 25, 2026 by PratikDhanave Loading…
Fix malformed 'pytlint' lint pragma -> 'pylint' in hd configs
#756 opened Jul 25, 2026 by PratikDhanave Loading…
Fix stale method reference in nn comments (_include_vision_embeddings -> _merge_mm_embeddings)
#755 opened Jul 25, 2026 by PratikDhanave Loading…
Fix load_params error reporting expected instead of actual checkpoint type
#754 opened Jul 25, 2026 by PratikDhanave Loading…
Fix non-f-string ValueError in t5gemma _normalize_token
#753 opened Jul 25, 2026 by PratikDhanave Loading…
Fix end-of-turn token leaking into streamed chat output
#747 opened Jul 25, 2026 by PratikDhanave Loading…
Fix incorrect batch-dim shape annotations on vmapped _get_new_text_tokens_inner
#746 opened Jul 25, 2026 by PratikDhanave Loading…
1
Fix wrong parameter name in diffusion attention-mask error message
#745 opened Jul 25, 2026 by PratikDhanave Loading…
1
Remove phantom label_smoothing attribute from NpoLoss docstring
#744 opened Jul 25, 2026 by PratikDhanave Loading…
1
Fix misleading custom-token id range in tokenizer error message
#743 opened Jul 25, 2026 by PratikDhanave Loading…
1
Fix broken file links in hackable_diffusion_adapter AGENT.md
#739 opened Jul 22, 2026 by ZayanKhan-12 Loading…
2
Gemma4: Support restoring checkpoints with custom dtype and fix precision issues.
#737 opened Jul 21, 2026 by copybara-service Bot Loading…
1
Fix ChatSampler.conversation crash for non-Gemma4 tokenizers (breaks ToolSampler multi-turn)
#734 opened Jul 17, 2026 by sohumt123 Loading…
1
Reverses the order of operations in Embedder.encode_audio to match the upstream Gemax training configuration (Norm -> Projection, instead of Projection -> Norm).
#730 opened Jul 15, 2026 by copybara-service Bot Loading…
Code update
#726 opened Jul 14, 2026 by copybara-service Bot Loading…
1
docs: add community health files and README updates
#724 opened Jul 9, 2026 by Mukller Loading…
1
Fix missing f-prefix in _normalize_token error message
#722 opened Jul 9, 2026 by HumphreySun98 Loading…
1
3
Use integer division for GQA reshape group size
#721 opened Jul 8, 2026 by HumphreySun98 Loading…
1
3
tests: relax RMSNorm tolerance to fix GPU precision failures
#718 opened Jul 5, 2026 by AbhishekGiri04 Loading…
1
fix: replace bare asserts with ValueError in _normalize_response (tool response parsing)
#716 opened Jul 5, 2026 by DhruvTilva Loading…
2
ProTip! What’s not been updated in a month: updated:<2026-06-30.