ms-swift: Train 600+ LLMs
Train large language models with ms-swift, a tool for startup founders, with 14k+ GitHub stars.
14,372 stars1,450 forksPythonUpdated 6/2/2026100% free ยท open source Train and fine-tune large language models using ms-swift, allowing startup founders to leverage AI capabilities for various applications
- โขWhen you need to customize a language model for your specific business needs
- โขWhen you want to integrate AI-powered text generation or understanding into your product
- โขWhen you're looking to improve the performance of a pre-trained language model on a particular task
- 1Install ms-swift using pip: `pip install ms-swift`
- 2Choose a pre-trained language model from the ms-swift repository, such as Qwen3.6 or Llama4
- 3Use the PEFT or Full-parameter tuning method to adapt the model to your specific task or dataset
- 4Evaluate the performance of the fine-tuned model using metrics such as accuracy or perplexity
Ready-to-paste prompt python -m ms_swift.train --model-name Qwen3.6 --task-name sentiment-analysis --dataset your_custom_dataset
Saves to your device
Topics
deepseek-r1
embedding
grpo
internvl
liger
llama
llama4
llm
lora
megatron
moe
multimodal
open-r1
peft
qwen3
qwen3-6
qwen3-omni
qwen3-vl
reranker
sft
What's inside โ free to inspect
No purchase needed
Read the entire source before you build โ unlike paid marketplaces that hide it behind a buy button.