Lambent/Mira-v1.28-dpo

Full fine-tune vs Lambent/Mira-v1.28-wave. Fine-tune: 100% of the weights changed, across 62 of 62 layers.

Tensors changed: Embeddings 2/4, Norms 341/484, MLP 275/294, Attention 427/464, other 1/1

2 derivatives checked

On the Hub · Downloads on Model Pulse