skills.rest
1thirteeng3
1thirteeng3COMMUNITY
Explore all skills by this author →
GitHub

slime-rl-training

Guide RL-based post-training of GLM models using slime with Megatron-LM integration.

Updated Mar 24, 2026
Search Tags:#mlops#llm#slime#rl-training#sglang#post-training#megatron-lm
One-click install
npx skills add https://github.com/1thirteeng3/greenmoire --skill slime-rl-training-1thirteeng3
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: slime-rl-training
Source: https://github.com/1thirteeng3/greenmoire/tree/main/integrations/hermes-agent/skills/mlops/training/slime
Command: npx skills add https://github.com/1thirteeng3/greenmoire --skill slime-rl-training-1thirteeng3

SYSTEM DOCUMENTATION & REQUIREMENTS

💡 This Skill includes scripts (resource) and references (resource) components.

What problem does it solve?

slime-rl-training provides guidance for LLM post-training with RL using slime, a Megatron+SGLang framework. Use when training GLM models, implementing custom data generation workflows, or needing tight Megatron-LM integration for RL scaling.

Core Features & Use Cases

  • Guides ML engineers through RL-based post-training of LLMs using slime.
  • Supports Megatron-LM and SGLang-based rollout and data buffering for GLM-family models.
  • Includes example workflows (GRPO training, asynchronous training, multi-turn agentic training) and pre-configured model scripts.

Quick Start

Configure and run a slime-driven RL post-training workflow to fine-tune a GLM model using Megatron-LM and SGLang.

Frequently Asked Questions about slime-rl-training

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

FAQPage Schema
How do I use RL post-training to fine-tune a GLM model with Megatron-LM?▼

RL post-training for GLM models requires tight Megatron-LM integration for RL scaling. This process configures slime-driven workflows to conduct SGLang-based rollout and data buffering, providing end-to-end guidance from setup to GRPO training execution.

Can I implement asynchronous training and multi-turn agentic workflows for LLMs?▼

Asynchronous training and multi-turn agentic workflows for LLMs are supported through pre-configured example workflows. These configurations leverage the slime framework to orchestrate complex rollout and data generation pipelines using SGLang and Megatron-LM.

What is RL-driven post-training for LLMs and when do I need it?▼

RL-driven post-training for LLMs applies reinforcement learning to fine-tune models after initial pre-training. You need it when implementing custom data generation workflows or scaling GLM-family model training using the Megatron+SGLang slime framework.

Does slime support custom data generation workflows for GLM-family models?▼

Slime fully supports custom data generation workflows for GLM-family models. It utilizes SGLang-based rollout and data buffering mechanisms to feed the RL training loop, enabling flexible pipeline construction for Megatron-LM scaling.

What is the best way to scale RL training for large language models?▼

Scaling RL training for large language models is best achieved through tight Megatron-LM integration. The slime framework provides pre-configured model scripts and asynchronous training setups to efficiently manage distributed SGLang rollouts and data buffering.

Related Skills

Explore More Software Engineering Skills→
blueskies1818blueskies1818
1

slime-rl-training

Guide LLM post-training with RL using Megatron-LM and SGLang.

Community
Advanced
Chris-Chai-MinjaeChris-Chai-Minjae

slime-rl-training

Guide RL post-training for large language models with slime.

Community
Advanced
KarlinskySKarlinskyS

slime-rl-training

Coordinate Megatron-LM training with SGLang rollouts for RL-based post-training.

Community
Advanced
Tommy-ywTommy-yw
539

slime-rl-training

Optimize large language models with Megatron-LM training and SGLang rollouts.

Community
Advanced