mradermacher/HunYuanOCR-SFT-GGUF

Same weights, converted (fewer layers) vs jaqja/HunYuanOCR-SFT. mradermacher/HunYuanOCR-SFT-GGUF matches part of jaqja/HunYuanOCR-SFT (266 of its 713 tensors; the rest of jaqja/HunYuanOCR-SFT has no counterpart). On the shared part: Same weights stored in another dtype (BF16 → F16/F32).

Tensors changed: Embeddings 0/1, Norms 0/97, MLP 0/72, Attention 0/96

On the Hub · Downloads on Model Pulse