mrMEtr/functiongemma-270m-it-mobile-actions

Partial fine-tune vs google/functiongemma-270m-it. Partial change: 37% of the weights changed (attention, mlp, norms), in 18 of 18 layers.

Tensors changed: Embeddings 0/1, Norms 30/109, MLP 54/54, Attention 72/72

On the Hub · Downloads on Model Pulse