texdata/Sokol-4B-SLO
Full fine-tune vs SicariusSicariiStuff/Qwen3.5-4B_Abliterated. Fine-tune: 93% of the weights changed, across 32 of 32 layers.
Tensors changed: Embeddings 1/4, Norms 103/203, Attention 214/320, MLP 96/196
Full fine-tune vs SicariusSicariiStuff/Qwen3.5-4B_Abliterated. Fine-tune: 93% of the weights changed, across 32 of 32 layers.
Tensors changed: Embeddings 1/4, Norms 103/203, Attention 214/320, MLP 96/196