blascotobasco/LFM2-24B-A2B-CinderWatch

Partial fine-tune vs huihui-ai/Huihui-LFM2-24B-A2B-abliterated. Partial change: 2% of the weights changed (attention, mlp, norms, other), in 19 of 40 layers.

Tensors changed: MLP 32/2768, Embeddings 0/2, other 6/90, Norms 1/100, Attention 4/40

2 derivatives checked