inclusionAI/Ling-3.0-flash-Fin

Full fine-tune vs inclusionAI/Ling-3.0-flash. Fine-tune: 100% of the weights changed, across 43 of 43 layers.

Tensors changed: Attention 460/460, MLP 2404/2404, lm_head 1/1, Norms 133/133, other 1/1, Embeddings 1/1

3 derivatives checked