From the model card
Fields below are copied from the tags and counters on the HuggingFace repository deepseek-ai/DeepSeek-R1 at our last fetch. They are set by the uploader, not verified by us; rows with no tag are omitted. How this page is made.
- Publisher (HF namespace)
- deepseek-ai
- Pipeline tag
- text-generation
- Library
- Transformers
- Weight formats
- safetensors
- License tag
mit— read the license file in the repo before relying on it- Papers cited
- arXiv:2501.12948
- Downloads (HF counter at last fetch)
- 1,013,158
- Likes (HF counter at last fetch)
- 13,608
- Model card
- https://huggingface.co/deepseek-ai/DeepSeek-R1
Use cases
- Mathematical problem solving requiring step-by-step derivation
- Code generation and debugging with transparent reasoning traces
- Logic and planning tasks where intermediate reasoning steps improve correctness
- Research benchmarking of reasoning-tuned open-weight models
Pros
- MIT license allows unrestricted commercial and research use
- Chain-of-thought output makes reasoning auditable and inspectable
- Competitive with proprietary models on MATH and competitive coding benchmarks
Cons
- 671B total weights require a multi-node cluster for full-precision inference
- Chain-of-thought verbosity inflates token usage and increases generation latency significantly
- Custom deepseek_v3 architecture requires non-standard loading code outside standard transformers
Tags
transformerssafetensorsdeepseek_v3text-generationconversationalcustom_codearxiv:2501.12948license:miteval-resultstext-generation-inferenceendpoints_compatiblefp8deploy:sagemakerregion:us