Retour à Deepseek R1 Distill
DeepSeek's R1 reasoning distilled into Llama 3 8B — strong chain-of-thought on a Llama base for broad ecosystem compatibility.
164K tokensGratuit / Poids ouvertsTransformerMIT
Benchmarks
GAIA
DeepSeek-R1-Distill-Llama-8B
27.6%DeepSeek-R1-Distill-Qwen-32B
57.5%DeepSeek-R1-Distill-Qwen-7B
25.6%TapeAgent v0.1 (GPT-4o) referencing distill
