Kindadodgy/qwen38-256k-on-16gb
Quantization vs unsloth/Qwen3.8-27B-GGUF. Kindadodgy/qwen38-256k-on-16gb is a GGUF quantization (BF16) of the same weights: 408 tensors match exactly and 458 within quantization error.
Tensors changed: Norms 0/216, other 0/290, Attention 0/164, MLP 0/195, lm_head 0/1