puwaer/DeepSeek-V4-Flash-0731-reap-150b
Partial fine-tune (fewer layers) vs deepseek-ai/DeepSeek-V4-Flash-0731. puwaer/DeepSeek-V4-Flash-0731-reap-150b matches part of deepseek-ai/DeepSeek-V4-Flash-0731 (35537 of its 72317 tensors; the rest of deepseek-ai/DeepSeek-V4-Flash-0731 has no counterpart). On the shared part: Partial change: 15% of the weights changed (mlp), in 43 of 43 layers.
Tensors changed: Embeddings 0/1, MLP 652/1044, Attention 0/851, Norms 0/235, other 0/4
2 derivatives checked
- puwaer/DeepSeek-V4-Flash-0731-reap-150b-gguf: Not comparable yet
- Tinker-Stack/DeepSeek-V4-Flash-0731-REAP-150b-TQ3_4S-GGUF: Not comparable yet