3.4 KiB
Setup
Use the repository's existing local .venv for Python. All Docker-backed checks and
experiments use the x86-64 daemon at ssh://[email protected] through
scripts/remote_docker.py; see remote-docker.md. Source, Python,
credentials, logs, and results stay on this workstation. Do not copy the repository or
create a Python environment on the Docker host.
uv sync
uv pip install -e /path/to/impossiblebench --no-deps
Why --no-deps
It is required, not a shortcut. ImpossibleBench declares
inspect_evals[swe_bench] @ git+https://github.com/UKGovernmentBEIS/inspect_evals from
unpinned git main, which forces huggingface_hub up to 1.2+, and swebench>=4.0.0, which
drags modal, GitPython, typer and pre-commit in as runtime dependencies.
The LiveCodeBench path we use imports none of it. Every swebench import in that package is
lazy and inside a function, and inspect_evals is declared but never imported.
--no-deps also skips datasets, which hf_dataset genuinely does need, so this repo
declares that one itself in pyproject.toml.
The staged SWE-bench path is intentionally separate. Its pinned optional dependencies and free nochange/oracle gate are documented in swe-validation.md; do not install them for LiveCodeBench-only work.
Two things to know. uv sync prunes ImpossibleBench, because it is not declared in
pyproject.toml, so rerun the install line after every sync. just install does both.
And ImpossibleBench mutates sys.path at import time to paper over a broken import, which
loads some modules twice, so do not trust module identity inside it.
The working directory
compose.yaml at the repo root sets working_dir: /workspace, which is a deliberate
deviation from the published benchmark. The file itself explains why.
Inspect looks for compose.yaml in the process working directory, not next to the task
definition, so every command has to run from the repo root. If you run from somewhere else
you silently get ImpossibleBench's own compose file, and the working directory reverts to
/. The justfile recipes handle this.
Checks that cost nothing
just smoke # fake model locally; task container on the remote Docker daemon
just test # offline unit tests, no network and no provider beyond mockllm
just smoke should end with status: success and a real score. If the tracebacks in that
score mention /workspace/test.py, the working directory override is live.
Spending money
.env holds OPENROUTER_API_KEY and is gitignored. The key carries a $2 cap as a fuse.
For a separate Kimi K3 or DeepSeek SWE population using ClinePass, put
CLINE_API_KEY in the same local .env. The SWE runner pins Cline's
https://api.cline.bot/api/v1 endpoint. Use one of the exact Inspect model IDs
openai-api/cline/cline-pass/kimi-k3,
openai-api/cline/cline-pass/deepseek-v4-pro, or
openai-api/cline/cline-pass/deepseek-v4-flash in a new frozen plan. GLM-5.3 Flash
stays on OpenRouter; do not mix the providers in one plan. ClinePass account quota
is not available from the documented API, so its account snapshots are marked
unavailable rather than using the OpenRouter balance. New ClinePass plans omit
the unverified OpenRouter reasoning settings; provider compatibility has not
been checked with a paid call.
Watch per-request prompt_n on any run longer than a few turns. Cache reuse fails silently
and costs 8 to 18 times as much without emitting an error.