0xSero/DeepSeek-V4-Flash-162B
Partial fine-tune (fewer layers) vs deepseek-ai/DeepSeek-V4-Flash. 0xSero/DeepSeek-V4-Flash-162B matches part of deepseek-ai/DeepSeek-V4-Flash (39534 of its 69187 tensors; the rest of deepseek-ai/DeepSeek-V4-Flash has no counterpart). On the shared part: Partial change: 17% of the weights changed (mlp), in 43 of 43 layers.
Tensors changed: Embeddings 0/1, MLP 788/1189, Attention 0/865, Norms 0/242, other 0/11
2 derivatives checked
- xik94/DeepSeek-V4-Flash-162B-REAP-GGUF: Not comparable yet
- 0xSero/DeepSeek-V4-Flash-162B-GGUF: Not comparable yet