skills.rest
Chris-Chai-Minjae
Chris-Chai-MinjaeCOMMUNITY
Explore all skills by this author →
GitHub

slime-rl-training

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

Updated Apr 23, 2026
Search Tags:#data-generation#glm#reinforcement-learning#rl#sglang#post-training#megatron-lm
One-click install
npx skills add https://github.com/Chris-Chai-Minjae/hermes-agent-r1-bridge --skill slime-rl-training-chris-chai-minjae
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: slime-rl-training
Source: https://github.com/Chris-Chai-Minjae/hermes-agent-r1-bridge/tree/main/optional-skills/mlops/slime
Command: npx skills add https://github.com/Chris-Chai-Minjae/hermes-agent-r1-bridge --skill slime-rl-training-chris-chai-minjae

SYSTEM DOCUMENTATION & REQUIREMENTS

💡 This Skill requires sglang-router, ray, torch, transformers.

What problem does it solve?

RL post-training for large language models using slime addresses the need for scalable, reproducible fine-tuning with reinforcement learning, tying Megatron-LM training and SGLang rollout together in a cohesive workflow.

Core Features & Use Cases

  • End-to-end RL post-training for GLM models with Megatron-LM and SGLang integration.
  • Custom data generation workflows and rollout management tailored to RL objectives.
  • Use cases include refining GLMs for instruction following, safety, and alignment, as well as research experiments exploring RL in large-scale language models.

Quick Start

Install slime and run the training workflow with your model checkpoint and prompt data to start RL post-training.

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 reinforcement learning post-training for large language models with Megatron-LM and SGLang?▼

Reinforcement learning post-training for large language models is guided by integrating Megatron-LM for training and SGLang for rollout into a cohesive, reproducible workflow using slime. You install slime, then run the training workflow with your model checkpoint and prompt data.

Can I use slime for custom data generation workflows in RL post-training?▼

Yes, custom data generation workflows are supported during RL post-training. The slime framework provides configurable data pipelines and rollout management tailored to specific RL objectives for GLM-scale models.

What dependencies do I need to set up before starting RL post-training with slime?▼

To start RL post-training with slime, you need to install sglang-router, ray, torch, and transformers. These dependencies satisfy the production-ready RL training loop requirements for scalable fine-tuning.

Does slime support RL fine-tuning for instruction following and safety alignment in GLM models?▼

Yes, slime supports RL fine-tuning for GLM models to improve instruction following, safety, and alignment. It enables end-to-end post-training tailored to these specific objectives and research experiments.

What is the best way to ensure reproducible experiments during large-scale RL post-training?▼

To ensure reproducible experiments during large-scale RL post-training, slime provides a production-ready training loop with configurable data pipelines. This ties Megatron-LM training and SGLang rollout together in a cohesive workflow.

Related Skills

Explore More Education & Research Skills→
richardnguyen0715richardnguyen0715

slime-rl-training

Train LLMs with reinforcement learning using Megatron-LM and SGLang.

Community
Advanced
hhhi21ghhhi21g

simpo-training

Optimizes language models for alignment and performance without a reference model.

Community
Advanced
hhhi21ghhhi21g

slime-rl-training

Scale RL post-training for LLMs with Megatron-LM and SGLang.

Community
Advanced
ovachieverovachiever
52

grpo-rl-training

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

Community
Advanced