meshllm/GLM-5.3-Flash-UD-Q4_K_XL-layers
Quantization (fewer layers) vs unsloth/GLM-5.3-Flash-GGUF. meshllm/GLM-5.3-Flash-UD-Q4_K_XL-layers matches part of unsloth/GLM-5.3-Flash-GGUF (26 of its 1412 tensors; the rest of unsloth/GLM-5.3-Flash-GGUF has no counterpart). On the shared part: meshllm/GLM-5.3-Flash-UD-Q4_K_XL-layers is a GGUF quantization (Q8_0/F32) of the same weights: 12 tensors match exactly and 14 within quantization error.
Tensors changed: Attention 0/7, MLP 0/6, Norms 0/3, other 0/10