prithivMLmods/GN-108036-Qwen-14B

Full fine-tune vs prithivMLmods/Calcium-Opus-14B-Elite3. Fine-tune: 100% of the weights changed, across 48 of 48 layers.

Tensors changed: Embeddings 1/1, Norms 97/97, MLP 144/144, Attention 336/336, lm_head 1/1

1 derivatives checked