# PRIME-RL/PRIME

Scalable RL solution for advanced reasoning of language models

Repository: https://github.com/PRIME-RL/PRIME
Canonical: https://ross.abutalabs.com/products/prime
Language: Python
License: Apache-2.0
License Family: permissive
Topics: llm, reasoning, rl
Last push: 2025-03-18T04:02:27+00:00

## Health v2 (maintenance only)
Score: 26/100 (v2, computed 2026-09-02T17:46:02.011165+00:00)
- activity 12, release rhythm 35, longevity 43
- inputs: {"age_days": 610, "days_push": 533, "days_rel": null, "gap_med": null, "n_releases_24m": 0}
- flags: no_releases
- formula: round(0.45*activity + 0.35*rhythm + 0.20*longevity); archived -> min(score, 10)

## Adoption (not part of the score)
Stars 1871, forks 116 (observed 2026-08-28T04:05:46.933898+00:00)

## What it is
PRIME (Process Reinforcement through Implicit Rewards) is an open-source Python framework for online reinforcement learning with process rewards to improve language model reasoning. It provides training, data preprocessing, and evaluation code, and has been integrated into the veRL framework.

## Use cases
- train language models with reinforcement learning for math reasoning
- run online RL with process rewards on LLMs
- reproduce PRIME paper results on reasoning benchmarks
- preprocess math datasets for RL training
- train implicit process reward models
- improve LLM reasoning beyond distillation

## When to choose
- you want to apply process-based RL to improve LLM reasoning
- you need an open-source, scalable RL training pipeline for language models
- you want to reproduce or build on the PRIME research results

## When to avoid
- you only need inference or serving of LLMs without training
- you need a simple supervised fine-tuning pipeline without RL
- you lack multi-GPU infrastructure for large-scale training

## Facets
- artifact type: library
- maturity: active
- function: llm-training, reinforcement-learning, machine-learning
- domain: large-language-models, machine-learning, artificial-intelligence
- platform: python
- tags: rlhf, process-reward-model, reasoning, implicit-rewards, verl, gpu, linux

## Member repositories
- PRIME-RL/PRIME (main) score 26

## Provenance
- Observed fields: from GitHub, fetched 2026-08-28T04:05:46.933898+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 2026-08-30T03:15:00.928012+00:00, confidence not recorded.
  - readme: https://github.com/PRIME-RL/PRIME (fetched 2026-08-28T04:05:46.933898+00:00, sha f56686a50f1b)
- Data as of 2026-08-30T08:39:29.467469+00:00.