usermma/LFM2.5-1.2B-MEGABRAIN-Thinking-Claude-Polaris-Deepseek-GLM-abliterated-failspy
Partial fine-tune vs DavidAU/LFM2.5-1.2B-MEGABRAIN-Thinking-Claude-Polaris-Deepseek-GLM. Partial change: 13% of the weights changed (attention, embeddings), in 2 of 16 layers.
Tensors changed: Embeddings 1/2, other 0/30, MLP 0/48, Norms 0/44, Attention 8/24