inclusionAI/Ming-Lite-Omni-1.5
Full fine-tune vs inclusionAI/Ming-Lite-Omni. Fine-tune: 100% of the weights changed, across 32 of 32 layers.
Tensors changed: Attention 640/640, MLP 1509/1509, Norms 167/173, Embeddings 3/3, other 14/14, lm_head 2/2
Full fine-tune vs inclusionAI/Ming-Lite-Omni. Fine-tune: 100% of the weights changed, across 32 of 32 layers.
Tensors changed: Attention 640/640, MLP 1509/1509, Norms 167/173, Embeddings 3/3, other 14/14, lm_head 2/2