DeepSeek-R1-Distill-Qwen 1.5B
A compact Qwen-based checkpoint distilled from DeepSeek-R1. DeepSeek publishes the distilled weights under MIT while preserving the base model’s applicable terms.
Demos
- Try distilled reasoning locally DeepSeek-R1-Distill-Qwen 1.5B by Hugging Face Transformers.js team.
Models
- DeepSeek-R1-Distill-Qwen 1.5B DeepSeek.