An agent skill and Python worker that let your coding agent experiment on an open language model: load it once, edit its weights reversibly, measure the effect, and export the result. It is inspired by weight-space programming (Takehi, Teoh, Zhang, and Khattab), in which a general-purpose LLM reads and writes a network's weights directly instead of relying on an optimizer such as gradient descent. Every snippet's source, output, and forward-pass count is saved. Version 0.5 is experimental, and snippets run as trusted Python with your filesystem and network access.
Tested on CPU, Apple MPS, NVIDIA H200, a 72B model split across eight NVIDIA H100s,
and SkyPilot with Slurm. Quantized loading, editing, saved bundles, and export are checked
by tests/live/validate_quantized.py on CUDA or MPS.
A deliberately silly weight edit, with exact undo. Try the example.
Requires Linux or macOS, git, and uv.
claude plugin marketplace add oumi-ai/weights-lab
claude plugin install oumi-weights-lab@oumi-weights-labThe plugin includes the oumi-weights-lab CLI. Its first run downloads PyTorch, so
allow a few minutes.
codex plugin marketplace add oumi-ai/weights-lab
codex plugin add oumi-weights-lab@oumi-weights-lab
uv tool install git+https://github.com/oumi-ai/weights-labCopy the whole skill folder, then install the CLI:
git clone https://github.com/oumi-ai/weights-lab
mkdir -p ~/.agents/skills
cp -R weights-lab/skills/oumi-weights-lab ~/.agents/skills/
uv tool install git+https://github.com/oumi-ai/weights-labThen ask your agent, for example: "Use oumi-weights-lab to check whether zeroing layer 3's MLP in Qwen/Qwen3-0.6B changes its answers to the questions in caps.jsonl."
To load LoRA adapters, Hugging Face datasets with lab.hf_rows, one model across
several GPUs (start --device-map balanced), or a 4-bit or 8-bit model
(start --quantize nf4), install the CLI as
uv tool install 'oumi-weights-lab[adapters,datasets,multi-gpu,quantized,saes] @ git+https://github.com/oumi-ai/weights-lab'.
The saes extra adds sparse autoencoders and transcoders (lab.load_sae) through
SAELens and TransformerLens, and is a larger install. The Claude Code plugin and remote
workers include it. To install only that one:
uv tool install 'oumi-weights-lab[saes] @ git+https://github.com/oumi-ai/weights-lab'.
Append @<tag> to a git+https:// URL to pin a release.
oumi-weights-lab start --session sessions/first --model HuggingFaceTB/SmolLM2-135M --device cpu
oumi-weights-lab send --session sessions/first --file edit.py --follow
oumi-weights-lab stop --session sessions/first
oumi-weights-lab backup --session sessions/first --output sessions/first.tar.gzWith edit.py:
rows = [{"prompt": "France's capital is", "options": [" Paris", " Rome"], "answer": 0}]
baseline = lab.evaluate(rows, name="baseline")
name = lab.parts("*mlp*weight")[0]["name"]
lab.zero(name)
edited = lab.evaluate(rows, name="zeroed")
print(lab.diff(baseline, edited))
lab.undo()
checkpoint = lab.export("restored") # saved under artifacts/Snippets share lab, model, tokenizer, Path, and artifacts, and their state
persists. Stdout is JSON; --follow streams snippet output to stderr.
From a source checkout, run the first two with uv run python -m examples.NAME --output-dir DIR.
examples/dogfood.pyruns baseline, edit, exact undo, export, stop, and backup on a tiny public GPT-2 through the CLI.examples/parity.pychecks GPT-2 generation and continuation scores against Hugging Face's own GPT-2 integration test.examples/replay_qwen_edit.pyreapplies the released weight-space programming edit of Qwen3.5-0.8B withlab.swap; see its docstring to run it.tests/live/validate_qwen_benchmark.pyevaluates it on MATH-500. On an H200 (500 problems, 4 samples each), the edit scored 51.25% against 7.85% for the base model, and samples reaching the 32,768-token cap fell from 92.90% to 30.55%. This reproduces the released edit, not its discovery. vLLM ran that benchmark;tests/live/validate_truncation.pychecks that the lab's own sampling hits the cap as often. On 40 of the problems without the presence penalty, which transformers lacks, both truncated 99.4% of the base model's samples, and the lab 51.3% of the edited model's against 50.6% for vLLM.