docs(examples): add Gemma4-31B SWE-bench eval steps + base-vs-SFT agentic result - #3585
Draft
athitten wants to merge 4 commits into
Draft
docs(examples): add Gemma4-31B SWE-bench eval steps + base-vs-SFT agentic result#3585athitten wants to merge 4 commits into
athitten wants to merge 4 commits into
Conversation
…ntic result Signed-off-by: Abhishree <athittenaman@nvidia.com> Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
Signed-off-by: Abhishree <athittenaman@nvidia.com> Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
Signed-off-by: Abhishree <athittenaman@nvidia.com> Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
Contributor
|
/ok to test 11c5071 |
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
2 participants
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
What
Documents Phase 3 (SWE-bench Verified eval) for the Gemma4-31B CoderForge CP-SFT recipe, and adds the OpenHands-3-tool eval scripts it references.
eval/: the 5 scripts the pipeline actually uses (setup, prewarm, serve+agent, grade) + the tool-call-emergence probe (oh3_run.py,enroot_env.py,grade_enroot.py,prewarm_images.py,probe_indist.py, and their.sub/.shlaunchers).Result
CoderForge SFT installs agentic behavior into the raw base
google/gemma-4-31B: the base emits 0/33 tool calls (never_toolcalled); the SFT model emits structuredtool_calls(think/execute_bash/str_replace_editor) on every turn.Validation
Ran the documented pipeline end-to-end on the step_799 checkpoint: setup ✓, serve+agent ✓ (vLLM served, 21/21 structured tool calls), grade ✓ (runs clean, enroot-errs 0). Scripts sanitized for public use (
/path/to+<account>placeholders, NVIDIA headers);ruff check/formatclean. Docs-only; no library code changed.🤖 Generated with Claude Code