p1atdev/dart-v2-moe-sft
Full fine-tune vs p1atdev/dart-v2-moe-base. Fine-tune: 100% of the weights changed, across 4 of 4 layers.
Tensors changed: lm_head 1/1, Embeddings 1/1, MLP 48/48, other 4/4, Norms 0/9, Attention 16/16
Full fine-tune vs p1atdev/dart-v2-moe-base. Fine-tune: 100% of the weights changed, across 4 of 4 layers.
Tensors changed: lm_head 1/1, Embeddings 1/1, MLP 48/48, other 4/4, Norms 0/9, Attention 16/16