ShareLab-SII/UniAR-RL

Partial fine-tune vs ShareLab-SII/UniAR-SFT. Partial change: 80% of the weights changed (attention, mlp, norms, other), in 36 of 36 layers.

Tensors changed: Embeddings 0/4, Norms 114/293, MLP 120/276, Attention 160/268, lm_head 0/1, other 1/2

On the Hub · Downloads on Model Pulse