AGENT SKILL

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.

License 许可
MIT
最近更新
2026-07-29
Stars
★ 0
Forks
0

01 前言 Front Matter

name
slime-rl-training
tags
[ "Reinforcement Learning", "Megatron-LM", "SGLang", "GRPO", "Post-Training", "GLM" ]
author
Orchestra Research
license
MIT
version
1.0.0
description
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.
dependencies
[ "sglang-router>=0.2.3", "ray", "torch>=2.0.0", "transformers>=4.40.0" ]

02 资源文件 2

03 SKILL.md 正文

slime: LLM Post-Training Framework for RL Scaling

slime is an LLM post-training framework from Tsinghua's THUDM team, powering GLM-4.5, GLM-4.6, and GLM-4.7. It connects Megatron-LM for training with SGLang for high-throughput rollout generation.

When to Use slime

Choose slime when you need:

  • Megatron-LM native training with SGLang inference
  • Custom data generation workflows with flexible data buffers
  • Training GLM, Qwen3, DeepSeek V3, or Llama 3 models
  • Research-grade framework with production backing (Z.ai)

Consider alternatives when:

  • You need enterprise-grade stability features → use miles
  • You want flexible backend swapping → use verl
  • You need PyTorch-native abstractions → use torchforge

Key Features

  • Training: Megatron-LM with full parallelism support (TP, PP, DP, SP)
  • Rollout: SGLang-based high-throughput generation with router
  • Data Buffer: Flexible prompt management and sample storage
  • Models: GLM-4.x, Qwen3, DeepSeek V3/R1, Llama 3

Architecture Overview

┌─────────────────────────────────────────────────────────┐
│                    Data Buffer                          │
│ - Prompt initialization and management                  │
│ - Custom data generation and filtering                  │
│ - Rollout sample storage                                │
└─────────────┬───────────────────────────┬───────────────┘
              │                           │
┌─────────────▼───────────┐ ┌─────────────▼───────────────┐
│ Training (Megatron-LM)  │ │ Rollout (SGLang + Router)   │
│ - Actor model training  │ │ - Response generation       │
│ - Critic (optional)     │ │ - Reward/verifier output    │
│ - Weight sync to rollout│ │ - Multi-turn support        │
└─────────────────────────┘ └─────────────────────────────┘

Installation

# Recommended: Docker
docker pull slimerl/slime:latest
docker run --rm --gpus all --ipc=host --shm-size=16g \
  -it slimerl/slime:latest /bin/bash

# Inside container
cd /root/slime && pip install -e . --no-deps

From Source

git clone https://github.com/THUDM/slime.git
cd slime
pip install -r requirements.txt
pip install -e .

Quick Start: GRPO Training

# Source model configuration
source scripts/models/qwen3-4B.sh

# Launch training
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 \
    ${MODEL_ARGS[@]} ${CKPT_ARGS[@]}

Workflow 1: Standard GRPO Training

Use this workflow for training reasoning models with group-relative advantages.

Prerequisites Checklist

  • [ ] Docker environment or Megatron-LM + SGLang installed
  • [ ] Model checkpoint (HuggingFace or Megatron format)
  • [ ] Training data in JSONL format

Step 1: Prepare Data

# data.jsonl format
{"prompt": "What is 2 + 2?", "label": "4"}
{"prompt": "Solve: 3x = 12", "label": "x = 4"}

Or with chat format:

{
    "prompt": [
        {"role": "system", "content": "You are a math tutor."},
        {"role": "user", "content": "What is 15 + 27?"}
    ],
    "label": "42"
}

Step 2: Configure Model

Choose a pre-configured model script:

# List available models
ls scripts/models/
# glm4-9B.sh, qwen3-4B.sh, qwen3-30B-A3B.sh, deepseek-v3.sh, llama3-8B.sh, ...

# Source your model
source scripts/models/qwen3-4B.sh

Step 3: Launch Training

python train.py \
    --actor-num-nodes 1 \
    --actor-num-gpus-per-node 8 \
    --rollout-num-gpus 8 \
    --advantage-estimator grpo \
    --use-kl-loss \
    --kl-loss-coef 0.001 \
    --prompt-data /path/to/train.jsonl \
    --input-key prompt \
    --label-key label \
    --apply-chat-template \
    --rollout-batch-size 32 \
    --n-samples-per-prompt 8 \
    --global-batch-size 256 \
    --num-rollout 3000 \
    --save-interval 100 \
    --eval-interval 50 \
    ${MODEL_ARGS[@]}

Step 4: Monitor Training

  • [ ] Check TensorBoard: tensorboard --logdir outputs/
  • [ ] Verify reward curves are increasing
  • [ ] Monitor GPU utilization across nodes

Workflow 2: Asynchronous Training

Use async mode for higher throughput by overlapping rollout and training.

When to Use Async

  • Large models with long generation times
  • High GPU idle time in synchronous mode
  • Sufficient memory for buffering

Launch Async Training

python train_async.py \
    --actor-num-nodes 1 \
    --actor-num-gpus-per-node 8 \
    --rollout-num-gpus 8 \
    --advantage-estimator grpo \
    --async-buffer-size 4 \
    --prompt-data /path/to/train.jsonl \
    ${MODEL_ARGS[@]}

Async-Specific Parameters

--async-buffer-size 4        # Number of rollouts to buffer
--update-weights-interval 2  # Sync weights every N rollouts

Workflow 3: Multi-Turn Agentic Training

Use this workflow for training agents with tool use or multi-step reasoning.

Prerequisites

  • [ ] Custom generate function for multi-turn logic
  • [ ] Tool/environment interface

Step 1: Define Custom Generate Function

# custom_generate.py
async def custom_generate(args, samples, evaluation=False):
    """Multi-turn generation with tool calling."""
    for sample in samples:
        conversation = sample.prompt

        for turn in range(args.max_turns):
            # Generate response
            response = await generate_single(conversation)

            # Check for tool call
            tool_call = extract_tool_call(response)
            if tool_call:
                tool_result = execute_tool(tool_call)
                conversation.append({"role": "assistant", "content": response})
                conversation.append({"role": "tool", "content": tool_result})
            else:
                break

        sample.response = response
        sample.reward = compute_reward(sample)

    return samples

Step 2: Launch with Custom Function

python train.py \
    --custom-generate-function-path custom_generate.py \
    --max-turns 5 \
    --prompt-data /path/to/agent_data.jsonl \
    ${MODEL_ARGS[@]}

See examples/search-r1/ for a complete multi-turn search example.


Configuration Reference

Three Argument Categories

slime uses three types of arguments:

1. Megatron Arguments (passed directly):

--tensor-model-parallel-size 2
--pipeline-model-parallel-size 1
--num-layers 32
--hidden-size 4096

2. SGLang Arguments (prefixed with --sglang-):

--sglang-mem-fraction-static 0.8
--sglang-context-length 8192
--sglang-log-level INFO

3. slime Arguments:

# Resource allocation
--actor-num-nodes 1
--actor-num-gpus-per-node 8
--rollout-num-gpus 8
--colocate  # Share GPUs between training/inference

# Data
--prompt-data /path/to/data.jsonl
--input-key prompt
--label-key label

# Training loop
--num-rollout 3000
--rollout-batch-size 32
--n-samples-per-prompt 8
--global-batch-size 256

# Algorithm
--advantage-estimator grpo  # or: gspo, ppo, reinforce_plus_plus
--use-kl-loss
--kl-loss-coef 0.001

Key Constraints

rollout_batch_size × n_samples_per_prompt = global_batch_size × num_steps_per_rollout

Example: 32 × 8 = 256 × 1


Data Buffer System

slime's data buffer enables flexible data management:

Basic Data Source

class RolloutDataSource:
    def get_samples(self, num_samples):
        """Fetch prompts from dataset."""
        return self.dataset.sample(num_samples)

    def add_samples(self, samples):
        """Called after generation (no-op by default)."""
        pass

Buffered Data Source (Off-Policy)

class RolloutDataSourceWithBuffer(RolloutDataSource):
    def __init__(self):
        self.buffer = []

    def add_samples(self, samples):
        """Store generated samples for reuse."""
        self.buffer.extend(samples)

    def buffer_filter(self, args, buffer, num_samples):
        """Custom selection logic (prioritized, stratified, etc.)."""
        return select_best(buffer, num_samples)

Common Issues and Solutions

Issue: SGLang Engine Crash

Symptoms: Inference engine dies mid-training

Solutions:

# Enable fault tolerance
--use-fault-tolerance

# Increase memory allocation
--sglang-mem-fraction-static 0.85

# Reduce batch size
--rollout-batch-size 16

Issue: Weight Sync Timeout

Symptoms: Training hangs after rollout

Solutions:

# Increase sync interval
--update-weights-interval 5

# Use colocated mode (no network transfer)
--colocate

Issue: OOM During Training

Symptoms: CUDA OOM in backward pass

Solutions:

# Enable gradient checkpointing
--recompute-activations

# Reduce micro-batch size
--micro-batch-size 1

# Enable sequence parallelism
--sequence-parallel

Issue: Slow Data Loading

Symptoms: GPU idle during data fetch

Solutions:

# Increase data workers
--num-data-workers 4

# Use streaming dataset
--streaming-data

Supported Models

Model Family Configurations
GLM GLM-4.5, GLM-4.6, GLM-4.7, GLM-Z1-9B
Qwen Qwen3 (4B, 8B, 30B-A3B), Qwen3-MoE, Qwen2.5
DeepSeek V3, V3.1, R1
Llama Llama 3 (8B, 70B)
Others Kimi K2, Moonlight-16B

Each model has pre-configured scripts in scripts/models/.


Advanced Topics

Co-location Mode

Share GPUs between training and inference to reduce memory:

python train.py \
    --colocate \
    --actor-num-gpus-per-node 8 \
    --sglang-mem-fraction-static 0.4 \
    ${MODEL_ARGS[@]}

Custom Reward Model

# custom_rm.py
class CustomRewardModel:
    def __init__(self, model_path):
        self.model = load_model(model_path)

    def compute_reward(self, prompts, responses):
        inputs = self.tokenize(prompts, responses)
        scores = self.model(inputs)
        return scores.tolist()
--custom-rm-path custom_rm.py

Evaluation Multi-Task

--eval-prompt-data aime /path/to/aime.jsonl \
--eval-prompt-data gsm8k /path/to/gsm8k.jsonl \
--n-samples-per-eval-prompt 16

Resources

04 同仓库更多 Skills

KrismithReddy12
KrismithReddy12/claude-code-templates

zapier-make-patterns

No-code automation democratizes workflow building. Zapier and Make (formerly Integromat) let non-developers automate business processes without writing code. But no-code doesn't mean no-complexity - these platforms have their own patterns, pitfalls, and breaking points. This skill covers when to use which platform, how to build reliable automations, and when to graduate to code-based solutions. Key insight: Zapier optimizes for simplicity and integrations (7000+ apps), Make optimizes for power

★ 0
KrismithReddy12
KrismithReddy12/claude-code-templates

yeet

Use only when the user explicitly asks to stage, commit, push, and open a GitHub pull request in one flow using the GitHub CLI (`gh`).

★ 0
KrismithReddy12
KrismithReddy12/claude-code-templates

workflow-automation

Workflow automation is the infrastructure that makes AI agents reliable. Without durable execution, a network hiccup during a 10-step payment flow means lost money and angry customers. With it, workflows resume exactly where they left off. This skill covers the platforms (n8n, Temporal, Inngest) and patterns (sequential, parallel, orchestrator-worker) that turn brittle scripts into production-grade automation. Key insight: The platforms make different tradeoffs. n8n optimizes for accessibility

★ 0
KrismithReddy12
KrismithReddy12/claude-code-templates

trigger-dev

Trigger.dev expert for background jobs, AI workflows, and reliable async execution with excellent developer experience and TypeScript-first design. Use when: trigger.dev, trigger dev, background task, ai background job, long running task.

★ 0
KrismithReddy12
KrismithReddy12/claude-code-templates

slack-automation

Automate Slack workspace operations including messaging, search, channel management, and reaction workflows through Composio's Slack toolkit.

★ 0
KrismithReddy12
KrismithReddy12/claude-code-templates

planning-with-files

Transforms workflow to use Manus-style persistent markdown files for planning, progress tracking, and knowledge storage. Use when starting complex tasks, multi-step projects, research tasks, or when the user mentions planning, organizing work, tracking progress, or wants structured output.

★ 0