# trevin-creator/autoresearch-mlx

Apple Silicon (MLX) port of Karpathy's autoresearch — autonomous AI research loops on Mac, no PyTorch required.

Repository: https://github.com/trevin-creator/autoresearch-mlx
Canonical: https://ross.abutalabs.com/products/autoresearch-mlx
Language: Python
License: MIT
License Family: permissive
Last push: 2026-07-02T13:49:49+00:00

## Health v2 (maintenance only)
Score: 55/100 (v2, computed 2026-09-02T17:46:02.011165+00:00)
- activity 90, release rhythm 35, longevity 12
- inputs: {"age_days": 178, "days_push": 62, "days_rel": null, "gap_med": null, "n_releases_24m": 0}
- flags: no_releases, young
- formula: round(0.45*activity + 0.35*rhythm + 0.20*longevity); archived -> min(score, 10)

## Adoption (not part of the score)
Stars 1813, forks 364 (observed 2026-08-28T04:05:39.714883+00:00)

## What it is
An Apple Silicon (MLX) port of Karpathy's autoresearch that runs autonomous AI research loops on a Mac without PyTorch or CUDA. A coding agent iteratively edits a single train.py, runs fixed 5-minute training experiments, and keeps or reverts changes based on validation bits-per-byte.

## Use cases
- run autonomous ML training experiments on Apple Silicon
- let a coding agent optimize a training loop automatically
- do LLM training research on a Mac without CUDA
- benchmark model configs under a fixed time budget
- explore hyperparameter search with keep-or-revert git workflow
- port PyTorch research loops to MLX

## When to choose
- you have an Apple Silicon Mac and no NVIDIA GPU
- you want agent-driven autonomous experimentation with MLX
- you want a minimal, transparent research loop controlled by a markdown protocol

## When to avoid
- you need PyTorch or CUDA ecosystem compatibility
- you need large-scale multi-GPU training
- you want a turnkey training pipeline rather than an experimental loop

## Facets
- artifact type: framework
- maturity: active
- function: llm-training, machine-learning, benchmarking, developer-tools
- domain: machine-learning, deep-learning, artificial-intelligence, developer-tools
- platform: python
- tags: mlx, apple-silicon, autonomous-research, agent-driven-experiments, neural-network-training, hyperparameter-search, karpathy-port, macos, gpu

## Member repositories
- trevin-creator/autoresearch-mlx (main) score 55

## Provenance
- Observed fields: from GitHub, fetched 2026-08-28T04:05:39.714883+00:00.
- Health v2: computed from the inputs above; adoption is never an input.
- Inferred fields (summary, facets, guidance): AI-extracted, prompt v1, taxonomy v1, on 2026-08-30T03:21:10.839276+00:00, confidence not recorded.
  - readme: https://github.com/trevin-creator/autoresearch-mlx (fetched 2026-08-28T04:05:39.714883+00:00, sha 578b86ede089)
- Data as of 2026-08-30T08:39:29.467469+00:00.