sayyidfareed/DeepSeek-V4.1-Flash-Next-DGX-Spark-512K

Same weights, repacked (fewer layers) vs deepseek-ai/DeepSeek-V4.1-Flash. sayyidfareed/DeepSeek-V4.1-Flash-Next-DGX-Spark-512K matches part of deepseek-ai/DeepSeek-V4.1-Flash (3925 of its 96085 tensors; the rest of deepseek-ai/DeepSeek-V4.1-Flash has no counterpart). On the shared part: Same weights in different files (re-saved, re-sharded or new metadata).

Tensors changed: other 0/20, Attention 0/765, MLP 0/611, Norms 0/248, Embeddings 0/8

On the Hub · Downloads on Model Pulse