Repository navigation
fix(mini_swe_agent_2): don't install agent deps into root venv (openai pin conflict) - #1916
Merged
ananthsub merged 1 commit intoJul 2, 2026
Conversation
…penai pin) The Quick Start told users to `uv pip install mini-swe-agent==2.1.0 swebench==4.1.0` into the root venv. mini-swe-agent allows a newer openai than nemo-gym pins (openai<=2.7.2), so that install upgrades the root venv's openai to 2.44.0. `gym env start` pins every per-server venv to the root venv's openai version via head_server_deps, which then conflicts with nemo-gym's pin and makes the mini_swe_agent_2 / policy_model venvs unsatisfiable, so `gym env start` fails with "Process `mini_swe_agent_2` finished unexpectedly". The root venv only needs `uv sync --extra dev --extra sandbox`; `gym env start` already builds the agent's own venv (with mini-swe-agent, swebench, and a compatible openai). Drop the manual install and document why not to add it. Signed-off-by: Ananth Subramaniam <ansubramania@nvidia.com>
ananthsub
added
the
r0.4.0
ananthsub
added a commit
that referenced
this pull request
Jul 2, 2026
…openai pin conflict) (1916)` into `r0.4.0` (#1917) beep boop [🤖]: Hi @ananthsub 👋, we've cherry picked #1916 into for you! 🚀 Please review and approve this cherry pick by your convenience! Signed-off-by: Ananth Subramaniam <ansubramania@nvidia.com> Signed-off-by: NeMo Bot <nemo-bot@nvidia.com> Co-authored-by: Ananth Subramaniam <ansubramania@nvidia.com>
This branch was previously deployed
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
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
Summary
Fixes a QA-blocking
gym env startfailure formini_swe_agent_2introduced by the Quick Start prerequisites.The Quick Start instructed installing the agent's runtime deps into the root venv:
mini-swe-agent==2.1.0allows any recentopenai(openai!=1.100.*), so this upgrades the root venv'sopenaifrom 2.7.2 → 2.44.0.gym env startthen pins every per-server venv to the root venv'sopenaiversion viahead_server_deps(openai=={openai_version}), which collides withnemo-gym'sopenai<=2.7.2pin in each sub-server venv:Fix
The root venv only needs
uv sync --extra dev --extra sandbox.gym env startalready builds this agent's own per-server venv from itsrequirements.txt(installingmini-swe-agent,swebench, and a compatibleopenai), so the manual root-venv install is both unnecessary and harmful. This PR drops it and documents why not to add it (plus how to recover: re-runuv sync).Immediate workaround for anyone hit by this
Test plan
uv sync --extra dev --extra sandboxalone →gym env startbuilds themini_swe_agent_2+ model server venvs successfully (verified previously end-to-end against OpenSandbox + a local vLLM).uv pip install mini-swe-agent==2.1.0upgradesopenai2.7.2 → 2.44.0 in the target venv (dry-run resolve), which is what triggers thehead_server_depsconflict.