0xSero/DeepSeek-V4-Flash-180B
Partial fine-tune (fewer layers) vs deepseek-ai/DeepSeek-V4-Flash. 0xSero/DeepSeek-V4-Flash-180B matches part of deepseek-ai/DeepSeek-V4-Flash (43758 of its 69187 tensors; the rest of deepseek-ai/DeepSeek-V4-Flash has no counterpart). On the shared part: Partial change: 18% of the weights changed (mlp), in 43 of 43 layers.
Tensors changed: Embeddings 0/1, MLP 874/1275, Attention 0/865, Norms 0/242, other 0/11
1 derivatives checked
- 0xSero/DeepSeek-V4-Flash-180B-GGUF: Not comparable yet