mradermacher/deepseek-math-10.8b-rl-GGUF

Same weights, converted vs Thermostatic/deepseek-math-10.8b-rl. Same weights stored in another dtype (F16 → F16/F32).

Tensors changed: lm_head 0/1, Norms 0/99, MLP 0/147, Attention 0/196, Embeddings 0/1

On the Hub · Downloads on Model Pulse