Skip to content

Add Lean Workbook tactics dataset#827

Open
derek-tml wants to merge 1 commit into
mainfrom
derek/lean-workbook-tactics-dataset
Open

Add Lean Workbook tactics dataset#827
derek-tml wants to merge 1 commit into
mainfrom
derek/lean-workbook-tactics-dataset

Conversation

@derek-tml

Copy link
Copy Markdown
Collaborator

Summary

  • Add a LeanWorkbookTacticsBuilder for internlm/Lean-Workbook proof-state next-tactic SFT examples.
  • Expose it through the chat SL CLI as base.dataset=lean_workbook_tactics.
  • Filter to proved rows with nonempty tactics and use a deterministic held-out split.

Validation

  • uv run --with pytest python -m pytest tinker_cookbook/recipes/chat_sl/chat_datasets_test.py
  • uv run --with ruff ruff check tinker_cookbook/recipes/chat_sl/chat_datasets.py tinker_cookbook/recipes/chat_sl/train.py tinker_cookbook/recipes/chat_sl/chat_datasets_test.py
  • uv run --with ruff ruff format --check tinker_cookbook/recipes/chat_sl/chat_datasets.py tinker_cookbook/recipes/chat_sl/train.py tinker_cookbook/recipes/chat_sl/chat_datasets_test.py

@derek-tml derek-tml marked this pull request as ready for review July 14, 2026 00:12
@derek-tml derek-tml force-pushed the derek/lean-workbook-tactics-dataset branch from d733122 to 92f10eb Compare July 14, 2026 00:17
@derek-tml derek-tml force-pushed the derek/lean-workbook-tactics-dataset branch from 92f10eb to d3e563a Compare July 14, 2026 00:19
@derek-tml derek-tml requested a review from dphuang2 July 14, 2026 00:56
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

1 participant