mradermacher/Qwen3.5-4B-MTP-Heretic-GGUF

Same weights, converted vs RBergBauer/Qwen3.5-4B-MTP-Heretic. Same weights stored in another dtype (F16/BF16 → F16/F32), plus 39 new tensors.

Tensors changed: Embeddings 0/1, Norms 0/81, MLP 0/96, Attention 0/32

On the Hub · Downloads on Model Pulse