ehristoforu/Gemma2-9b-it-train6-plus_layer-1

Partial fine-tune vs ehristoforu/Gemma2-9b-it-train6. Partial change: 87% of the weights changed (attention, mlp, norms), in 32 of 32 layers.

Tensors changed: Embeddings 0/1, Norms 128/129, MLP 96/96, Attention 128/128

On the Hub · Downloads on Model Pulse