giannisan/Hy3-ds4-gguf
Quantization vs tencent/Hy3. giannisan/Hy3-ds4-gguf is a GGUF quantization (IQ2_XXS/Q8_0) of the same weights: 407 tensors match exactly and 570 within quantization error.
Tensors changed: Attention 0/324, lm_head 0/1, MLP 0/323, other 0/1, Norms 0/327, Embeddings 0/1