gplsi/Aitana-TA-2B-S
Partial fine-tune vs BSC-LT/salamandraTA-2b-instruct. Partial change: 77% of the weights changed (attention, lm_head, mlp, norms), in 24 of 24 layers.
Tensors changed: lm_head 1/1, Embeddings 0/1, Norms 12/49, MLP 72/72, Attention 96/96
2 derivatives checked
- mradermacher/Aitana-TA-2B-S-i1-GGUF: Quantization
- mradermacher/Aitana-TA-2B-S-GGUF: Same weights, converted