[LLM] Add Terminal-Bench evaluations and configurable CI - #23382
Closed
mergennachin wants to merge 1 commit into
Closed
mergennachin wants to merge 1 commit into
mergennachin wants to merge 1 commit into
Conversation
Add setup and run commands under examples/llm_server/evals so developers can evaluate exported models through the existing LLM server. Terminal-Bench is the first harness, with pinned Harbor/mini-SWE-agent versions and task subsets, model settings under evals/configs/, separate context and output budgets, and recorded task rewards, token usage, timings, trajectories, and verifier evidence. Start review with evals/README.md and the example TOML, then terminal_bench/setup.sh and runner.py, followed by the CI workflow. Setup supports Ubuntu Docker Engine and macOS Colima; execution supports Docker Desktop too. Each trial owns a fresh server, with optional named-session affinity and resumable results. The periodic workflow uses configurable Linux GPU and macOS runners and stays disabled until LLM_SERVER_EVAL_TARGETS is provisioned. Validation: 36 evaluation tests passed, including the current Python HTTP launcher with anonymous and named sessions using a fixture worker. Lintrunner, ShellCheck, Bash syntax checks, Actionlint, and the local prerequisite check passed. Earlier real Qwen3-0.6B trials reached the verifier but failed tool-call formatting; they provide integration evidence, not a task-quality baseline. This standalone revision has not run a new model evaluation or dispatched periodic CI. Authored with assistance from OpenAI Codex.
🔗 Helpful Links🧪 See artifacts and rendered test results at hud.pytorch.org/pr/pytorch/executorch/23382
Note: Links to docs will display an error until the docs builds have been completed. ❌ 1 New Failure, 146 PendingAs of commit c735eaa with merge base 0b3d26d ( This comment was automatically generated by Dr. CI and updates every 15 minutes. |
meta-cla
Bot
added
the
CLA Signed
This PR needs a
|
This branch was successfully 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
1 participant
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.
Add setup and run commands under examples/llm_server/evals so developers can evaluate exported models through the existing LLM server. Terminal-Bench is the first harness, with pinned Harbor/mini-SWE-agent versions and task subsets, model settings under evals/configs/, separate context and output budgets, and recorded task rewards, token usage, timings, trajectories, and verifier evidence.
Start review with evals/README.md and the example TOML, then terminal_bench/setup.sh and runner.py, followed by the CI workflow. Setup supports Ubuntu Docker Engine and macOS Colima; execution supports Docker Desktop too. Each trial owns a fresh server, with optional named-session affinity and resumable results. The periodic workflow uses configurable Linux GPU and macOS runners and stays disabled until LLM_SERVER_EVAL_TARGETS is provisioned.
Validation: 36 evaluation tests passed, including the current Python HTTP launcher with anonymous and named sessions using a fixture worker. Lintrunner, ShellCheck, Bash syntax checks, Actionlint, and the local prerequisite check passed. Earlier real Qwen3-0.6B trials reached the verifier but failed tool-call formatting; they provide integration evidence, not a task-quality baseline. This standalone revision has not run a new model evaluation or dispatched periodic CI.
Authored with assistance from OpenAI Codex.