EasyR1: An Efficient, Scalable, Multi-Modality RL Training Framework based on veRL

搜索、推荐、广告、用增等工业界实践文章收集（来源：知乎、Datafuntalk、技术公众号）

A unified architecture deep learning framework designed specifically for ultra-large-scale sparse models.

HierarchicalKV is a part of NVIDIA Merlin and provides hierarchical key-value storage to meet RecSys requirements. The key capability of HierarchicalKV is to store key-value feature-embeddings on h…

Supercharge Your LLM with the Fastest KV Cache Layer

A Flexible Framework for Experiencing Cutting-edge LLM Inference Optimizations

Train speculative decoding models effortlessly and port them smoothly to SGLang serving.

Efficient and easy multi-instance LLM serving

Use PEFT or Full-parameter to CPT/SFT/DPO/GRPO 500+ LLMs (Qwen3, Qwen3-MoE, Llama4, GLM4.5, InternLM3, DeepSeek-R1, ...) and 200+ MLLMs (Qwen3-VL, Qwen3-Omni, InternVL3.5, Ovis2.5, Llava, GLM4v, Ph…

My learning notes/codes for ML SYS.

Universal memory layer for AI Agents

An Efficient and User-Friendly Scaling Library for Reinforcement Learning with Large Language Models

Train your Agent model via our easy and efficient framework

Lightning-Fast RL for LLM Reasoning and Agents. Made Simple & Flexible.

Super-Efficient RLHF Training of LLMs with Parameter Reallocation

An easy-to-use framework for large scale recommendation algorithms.

HLLM: Enhancing Sequential Recommendations via Hierarchical Large Language Models for Item and User Modeling

Pytorch domain library for recommendation systems

An Easy-to-use, Scalable and High-performance RLHF Framework based on Ray (PPO & GRPO & REINFORCE++ & vLLM & Ray & Dynamic Sampling & Async Agentic RL)

Reproduce R1 Zero on Logic Puzzle

Efficient Triton Kernels for LLM Training

verl: Volcano Engine Reinforcement Learning for LLMs

Train transformer language models with reinforcement learning.

Scalable toolkit for efficient model alignment

Unified Efficient Fine-Tuning of 100+ LLMs & VLMs (ACL 2024)

DeepEP: an efficient expert-parallel communication library

FlashMLA: Efficient Multi-head Latent Attention Kernels

SGLang is a fast serving framework for large language models and vision language models.

Let your Claude able to think