pdelobelle/DeepSeek-R1-GGUF-MTP

Quantized MTP head of Deepseek R1. For use with the Unsloth's Q4_K quants.

Llama.cpp does not support MTP heads, but vLLM does.