skills.rest
NITISH-gitbit
NITISH-gitbitCOMMUNITY
Explore all skills by this author →
GitHub

slime-rl-training

Streamline Megatron-LM reinforcement learning post-training with SGLang rollout generation.

Updated Jun 26, 2026
Search Tags:#scaling#large language models#SGLang#post-training#Megatron-LM#reforcement learning
One-click install
npx skills add https://github.com/NITISH-gitbit/hermes-custom --skill slime-rl-training-nitish-gitbit
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: slime-rl-training
Source: https://github.com/NITISH-gitbit/hermes-custom/tree/main/optional-skills/mlops/slime
Command: npx skills add https://github.com/NITISH-gitbit/hermes-custom --skill slime-rl-training-nitish-gitbit

SYSTEM DOCUMENTATION & REQUIREMENTS

💡 This Skill requires sglang-router>=0.2.3, ray, torch>=2.0.0, transformers>=4.40.0, and includes scripts (resource) and references (resource) and assets (resource) components.

What problem does it solve?

This Skill addresses the challenge of scaling reinforcement learning post-training with large language models like Megatron-LM, offering a framework for custom data generation workflows and tight integration with SGLang for high-throughput rollout generation.

Core Features & Use Cases

  • Megatron-LM Integration: Seamless connection with Megatron-LM for training and SGLang for high-throughput rollout generation.
  • Custom Data Generation: Flexible data buffers for custom prompt management and sample storage.
  • Use Case: Ideal for training GLM models, implementing custom data generation workflows, or achieving tight integration with Megatron-LM for RL scaling.

Quick Start

Run the following command to launch the training process: python train.py --actor-num-nodes 1 --actor-num-gpus-per-node 4 --rollout-num-gpus 4 --advantage-estimator grpo --use-kl-loss --kl-loss-coef 0.001 --rollout-batch-size 32 --n-samples-per-prompt 8 --global-batch-size 256 --num-rollout 3000 --prompt-data /path/to/data.jsonl

Frequently Asked Questions about slime-rl-training

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

FAQPage Schema
How do I scale reinforcement learning post-training with Megatron-LM?▼

Reinforcement learning post-training with Megatron-LM is scaled using the slime framework, which provides custom data generation workflows and integrates SGLang for high-throughput rollout generation to streamline the training process for large language models.

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

Before starting RL post-training with slime, you must install sglang-router (>=0.2.3), ray, torch (>=2.0.0), and transformers (>=4.40.0). You also need to prepare your prompt data in a JSONL file format for the training process.

Can I use SGLang for rollout generation in Megatron-LM reinforcement learning?▼

Yes, you can use SGLang for rollout generation in Megatron-LM reinforcement learning. The slime framework provides tight integration with SGLang to enable high-throughput rollout generation during the post-training phase.

How do I launch reinforcement learning training with GRPO advantage estimation?▼

To launch reinforcement learning training with GRPO advantage estimation, run the train.py script with parameters like --advantage-estimator grpo, --use-kl-loss, and --rollout-batch-size. You must also specify your actor and rollout GPU configurations.

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

Yes, slime supports custom data generation workflows for GLM models. It provides flexible data buffers for custom prompt management and sample storage, making it ideal for training GLM models with reinforcement learning.

Related Skills

Explore More Software Engineering Skills→
richardnguyen0715richardnguyen0715

slime-rl-training

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

Community
Advanced
ovachieverovachiever
52

grpo-rl-training

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

Community
Advanced
Tommy-ywTommy-yw
539

slime-rl-training

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

Community
Advanced
eyadsibaieyadsibai
7

llm-training

Compare frameworks and techniques for training and finetuning large language models.

Community
Intermediate