prem-research/DeepSeek-V4-Flash-0731-abliterated
LoRA-style vs deepseek-ai/DeepSeek-V4-Flash-0731. Looks like a merged LoRA (or a fine-tune with frozen norms): only the weight matrices of w2, wo_b changed, in 20 of 43 layers; norms, biases and embeddings are untouched.
Tensors changed: Embeddings 0/1, MLP 24/1055, Attention 19/642, Norms 0/249, other 0/11