autoresearch-ml
Autonomous LLM training optimization with GPU support. Runs 5-minute training experiments, measures val_bpb, keeps improvements or reverts — repeat forever. Use this skill when the user asks to "train a model autonomously", "optimize LLM training", "run ML experiments", "autoresearch with GPU", "optimize val_bpb", "autonomous ML training", "LLM pretraining loop", "setup ML autoresearch", "GPU training experiments", "pretrain from scratch", "speed up training", "lower my loss", "GPU optimization", "CUDA training", or mentions "train.py", "prepare.py", "bits per byte", "val_bpb", "NVIDIA GPU tra
npx skills add majiayu000/claude-skill-registry --skill autoresearch-ml --agent claude-code
Same command for any agent — swap --agent for codex, cursor, copilot.
Weekly change comes from our own snapshots, not the repository page — it measures attention, not adoption.
# Autoresearch ML: Autonomous LLM Training Optimization An autonomous experiment loop for single-GPU LLM pretraining. Edit `train.py` → commit → run 5-minute training → measure `val_bpb` → keep improvement or revert → **repeat forever**. This skill is self-contained — it includes everything needed to set up and run the loop. ## Setup Phase ### 1. Copy Template Assets Copy the bundled training template to the project directory: ```bash cp ${CLAUDE_SKILL_DIR}/assets/prepare.py . cp ${CLAUDE_SKILL_DIR}/assets/train.py . cp ${CLAUDE_SKILL_DIR}/assets/pyproject.toml . cp ${CLAUDE_SKILL_DIR}/assets/program.md . ``` ### 2. Install and Prepare ```bash uv sync # Install dependencies uv run prepare.py # Download data shards, train tokenizer (~2 min) ``` ### 3. Verify GPU ```bash nvidia-smi python -c "import torch; print(f'CUDA: {torch.cuda.is_available()}, Device: {torch.cuda.get_device_name()}, VRAM: {torch.cuda.get_device_properties(0).total_mem / 1e9:.1f} GB')" ``` ### 4. Initialize the Experiment Session 1. Create a branch: `git checkout -b autoresearch/<tag>-<date>` 2. Ensure session files are gitignored (critical — `git revert` will fail if tracked): ```bash echo -e "autoresearch.jsonl
- Setup Phase
- 1. Copy Template Assets
- 2. Install and Prepare
- 3. Verify GPU
- 4. Initialize the Experiment Session
- The Experiment Loop
- Decision Rules
- Simplicity Criterion
- Constraints
- Don't Thrash
- Handling User Messages
- Logging to autoresearch.jsonl
- ASI (Actionable Side Information)
- Resuming After Context Reset
cp ${CLAUDE_SKILL_DIR}/assets/prepare.py .
cp ${CLAUDE_SKILL_DIR}/assets/train.py .
cp ${CLAUDE_SKILL_DIR}/assets/pyproject.toml .
cp ${CLAUDE_SKILL_DIR}/assets/program.md .
uv sync # Install dependencies
uv run prepare.py # Download data shards, train tokenizer (~2 min)
nvidia-smi
echo -e "autoresearch.jsonl\nrun.log" >> .gitignore
git add .gitignore && git commit -m "autoresearch: add session files to gitignore"
bash ${CLAUDE_SKILL_DIR}/scripts/log-experiment.sh \What does the autoresearch-ml skill do?
Autonomous LLM training optimization with GPU support. Runs 5-minute training experiments, measures val_bpb, keeps improvements or reverts — repeat forever. Use this skill when the user asks to "train a model autonomously", "optimize LLM training", "run ML experiments", "autoresearch with GPU", "optimize val_bpb", "autonomous ML training", "LLM pretraining loop", "setup ML autoresearch", "GPU training experiments", "pretrain from scratch", "speed up training", "lower my loss", "GPU optimization", "CUDA training", or mentions "train.py", "prepare.py", "bits per byte", "val_bpb", "NVIDIA GPU tra
How do I install it?
Run `npx skills add majiayu000/claude-skill-registry --skill autoresearch-ml --agent claude-code` — it drops the skill into your project so the agent can pick it up. Swap the --agent value for codex, cursor or copilot if you use one of those.
Where does this skill come from?
From majiayu000/claude-skill-registry, a repository with 534 stars. We read it straight from the repository tree rather than a submitted listing, so what you see here is what is actually published.
Is a popular skill a good skill?
Not necessarily. Stars measure attention, not adoption — a repository can trend for a week and be abandoned. That is why we show the weekly change from our own snapshots next to the total, instead of a single flattering number.