Mimoniske/gemma-3-it-vl-40B-Gemini-Heretic-Uncensored-Thinking

Full fine-tune vs coder3101/gemma-3-27b-it-heretic. Fine-tune: 93% of the weights changed, across 62 of 62 layers.

Tensors changed: Embeddings 0/4, Norms 228/484, MLP 186/294, Attention 248/464, other 0/1

4 other repos hold byte-for-byte identical weight files.

On the Hub · Downloads on Model Pulse