ms-swift 开源项目推荐（9 个）

yanght27 /GPU-Perf-Playground

GPU 性能与 AI Infra 学习项目：CUDA/Triton 算子、NCU/NSYS、vLLM/SGLang/TRT-LLM/ms-swift、PyTorch/DeepSpeed/ms-swift 训练、并行架构

modelscope /mcore-bridge

MCore-Bridge: Providing Megatron-Core model definitions for state-of-the-art large models and making Megatron training as simple as Transformers — with support for 300+ large language models (Qwen3-Next, GLM-5.3, Deepseek-V4, MiniMax-2.7, ...) and 200+ multimodal large models (Qwen3.8, Qwen3-Omni, Gemma4, ...).

guanwei49 /EMIT

EMIT: Enhancing MLLMs for Industrial Anomaly Detection via Difficulty-Aware GRPO

nbadino /tabularium

State-of-the-art OCR and fine-tuning for historical documents, made simple for researchers. Model-agnostic, open-weight models, local or rented GPUs, multi-user.

jerry1993-tech /mimo-agent-grpo-swift

An end-to-end reproduction of Xiaomi MiMo-V2.6's open-source interactive Agent RL pipeline on MS-Swift — multi-turn tool use, sandboxed execution, real-execution verification, and GRPO. | 基于 MS-Swift 端到端复刻小米 MiMo-V2.6 开源的交互式 Agent RL 训练流程，覆盖多轮工具调用、隔离沙箱、真实执行验证与 GRPO 训练全链路。

jiacheng-zheng-LLM /Qwen3-8B-Fin-Posttrain

基于ms-swift的Qwen3_8B 金融推理两阶段后训练（LoRA SFT->GRPO）。

0KEAHA /lightning-launch

Agentic RL, ready to run. Bring your agent, customize rewards, train with ms-swift and Megatron.

vladimiralbrekhtccr /ms-swift-audio-integration

Audio-model training support in ms-swift (used for Oylan2)

taishan1994 /Concur

ConCuR: Conciseness Makes State-of-the-Art Kernel Generation (Reproduced)