gplsi/Aitana-2B-SI-Instruct-Aligned

Partial fine-tune vs gplsi/Aitana-2B-SI-Instruct. Partial change: 53% of the weights changed (attention, mlp, norms), in 24 of 24 layers.

Tensors changed: lm_head 0/1, Embeddings 0/1, Norms 10/49, MLP 72/72, Attention 95/96

1 derivatives checked

• mradermacher/Aitana-2B-SI-Instruct-Aligned-GGUF: Same weights, converted

On the Hub · Downloads on Model Pulse