meshllm/DeepSeek-V4-Flash-0731-UD-Q4_K_XL-layers
Full fine-tune (fewer layers) vs unsloth/DeepSeek-V4-Flash-0731-GGUF. meshllm/DeepSeek-V4-Flash-0731-UD-Q4_K_XL-layers matches part of unsloth/DeepSeek-V4-Flash-0731-GGUF (26 of its 81 tensors; the rest of unsloth/DeepSeek-V4-Flash-0731-GGUF has no counterpart). On the shared part: meshllm/DeepSeek-V4-Flash-0731-UD-Q4_K_XL-layers is a GGUF (MXFP4/Q8_0) of different weights. Fine-tune: 100% of the weights changed, across 0 of 0 layers.
Tensors changed: MLP 10/10, Attention 9/9, Norms 4/4, other 3/3