skills.rest
AlexiosBluffMara
AlexiosBluffMaraCOMMUNITY
Explore all skills by this author →
GitHub

slime-rl-training

Run GRPO RL post-training for GLM models with Megatron-LM and SGLang.

2|Updated Apr 25, 2026
Search Tags:#llm#reinforcement-learning#slime#rl#sglang#post-training#Megatron-LM
One-click install
npx skills add https://github.com/AlexiosBluffMara/mercury --skill slime-rl-training-alexiosbluffmara
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: slime-rl-training
Source: https://github.com/AlexiosBluffMara/mercury/tree/main/optional-skills/mlops/slime
Command: npx skills add https://github.com/AlexiosBluffMara/mercury --skill slime-rl-training-alexiosbluffmara

SYSTEM DOCUMENTATION & REQUIREMENTS

💡 This Skill includes references (resource) components.

What problem does it solve?

slime provides guidance for using the slime framework to perform reinforcement-learning-based post-training on large language models (LLMs), enabling structured, repeatable improvement loops.

Core Features & Use Cases

  • Guidance for integrating Megatron-LM with SGLang to support scalable RL post-training on GLM-family models.
  • Templates and workflows for automated data generation, rollout, and training loops, along with data buffering and evaluation hooks.
  • Use cases include research experiments, performance tuning, and production-ish workflows for RL-based model improvement.

Quick Start

Launch slime GRPO training workflow using Megatron-LM and SGLang to scale RL post-training for GLM models.

Frequently Asked Questions about slime-rl-training

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I run RL post-training for LLMs using Megatron-LM and SGLang?▼

You can perform RL post-training by launching the slime GRPO training workflow, which integrates Megatron-LM with SGLang to scale structured improvement loops for GLM-family models.

What is GRPO training and how does it work with slime for large language models?▼

GRPO training in slime provides a standardized loop for data generation, rollout, and training, allowing configurable prompts and multi-turn RL configurations for large language models.

Can I configure multi-turn RL workflows and data buffering for GLM models?▼

Yes, you can configure multi-turn RL workflows and manage data buffering using slime templates, which support automated data generation and evaluation hooks for GLM-family models.

Does slime support scalable reinforcement learning training for production environments?▼

slime supports scalable RL post-training by integrating Megatron-LM with SGLang, offering templates for research experiments and production-ish workflows with configurable resource management.

What are the limitations of using slime for reinforcement learning post-training?▼

slime is specifically tailored for GLM-family models using Megatron-LM and SGLang, meaning its standardized GRPO training workflows may not directly apply to other model architectures.

Related Skills

Explore More Software Engineering Skills→
CUexterCUexter
2

slime-rl-training

Run RL-based LLM post-training with Megatron-LM and SGLang for scalable rollout and GRPO workflows.

Community
Advanced
overviewlabsoverviewlabs

slime-rl-training

Guides RL post-training of large language models using Megatron-LM and SGLang integration.

Community
Advanced
ovachieverovachiever
52

grpo-rl-training

Apply GRPO/RL training with TRL to optimize language models.

Community
Advanced
HarriesHarries

slime-rl-training

Automate GRPO-style RL post-training workflows for large language models with Megatron-LM and SGLang.

Community
Advanced