# SamsungSAILMontreal/TinyRecursiveModels

Repository: https://github.com/SamsungSAILMontreal/TinyRecursiveModels
Canonical: https://ross.abutalabs.com/products/tinyrecursivemodels
Language: Python
License: MIT
License Family: permissive
Archived: true
Last push: 2026-04-01T00:50:39+00:00

## Health v2 (maintenance only)
Score: 10/100 (v2, computed 2026-09-03T02:20:16.233290+00:00)
- activity 75, release rhythm 35, longevity 23
- inputs: {"age_days": 330, "days_push": 155, "days_rel": null, "gap_med": null, "n_releases_24m": 0}
- flags: no_releases, archived
- formula: round(0.45*activity + 0.35*rhythm + 0.20*longevity); archived -> min(score, 10)

## Adoption (not part of the score)
Stars 6566, forks 1039 (observed 2026-08-28T04:09:45.262896+00:00)

## What it is
Official codebase for the Tiny Recursive Model (TRM) paper, a recursive reasoning approach where a tiny 7M-parameter neural network iteratively improves its answers on hard reasoning tasks like ARC-AGI. It provides training and evaluation code implemented in Python.

## Use cases
- train a tiny 7M parameter model on ARC-AGI puzzles
- reproduce recursive reasoning results from the TRM paper
- experiment with small networks that solve hard reasoning tasks without LLMs
- compare TRM against the Hierarchical Reasoning Model baseline
- run experiments on Sudoku-Extreme and other reasoning benchmarks

## When to choose
- you want to train or reproduce tiny recursive reasoning models from scratch
- you need a parameter-efficient alternative to large LLMs for structured reasoning tasks
- you are researching recursive reasoning or ARC-AGI benchmarks

## When to avoid
- you need a production-ready inference library or API
- you expect active maintenance - the repo has been archived as read-only
- you want a general-purpose LLM framework or pretrained foundation model

## Facets
- artifact type: library
- maturity: maintenance
- function: machine-learning, deep-learning, llm-training
- domain: machine-learning, deep-learning, artificial-intelligence
- platform: python
- tags: recursive-reasoning, tiny-networks, arc-agi, research-code, small-language-models, gpu, linux

## Member repositories
- SamsungSAILMontreal/TinyRecursiveModels (main) score 10

## Provenance
- Observed fields: from GitHub, fetched 2026-08-28T04:09:45.262896+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 2026-08-29T17:43:40.288763+00:00, confidence not recorded.
  - readme: https://github.com/SamsungSAILMontreal/TinyRecursiveModels (fetched 2026-08-28T04:09:45.262896+00:00, sha f9eb6625983f)
- Data as of 2026-08-30T08:39:29.467469+00:00.