primitive-ai/DeepSeek-V4-Flash-Vision-Exp-REAP-145B

Partial fine-tune (fewer layers) vs deepseek-ai/DeepSeek-V4-Flash-Vision-Exp. primitive-ai/DeepSeek-V4-Flash-Vision-Exp-REAP-145B matches part of deepseek-ai/DeepSeek-V4-Flash-Vision-Exp (34505 of its 72633 tensors; the rest of deepseek-ai/DeepSeek-V4-Flash-Vision-Exp has no counterpart). On the shared part: Partial change: 12% of the weights changed (mlp), in 43 of 43 layers.

Tensors changed: Embeddings 0/1, MLP 476/865, Attention 0/851, Norms 0/235, other 0/4

On the Hub · Downloads on Model Pulse