Artples/L-MChat-Small

Full fine-tune vs Weyaxi/Einstein-v4-phi2. Fine-tune: 97% of the weights changed, across 32 of 32 layers.

Tensors changed: Embeddings 1/1, Norms 53/66, MLP 124/128, Attention 248/256, lm_head 2/2

2 derivatives checked

• bunnycore/Phi-2-OmniMatrix: Full fine-tune
• bunnycore/Phi-2-Hyper: Full fine-tune

On the Hub · Downloads on Model Pulse