automerger/MeliodasM7-7B

Full fine-tune vs AurelPx/Meliodas-7b-dare. Fine-tune: 92% of the weights changed, across 31 of 32 layers.

Tensors changed: MLP 87/96, Attention 120/128, Norms 14/65, Embeddings 1/1, lm_head 1/1

1 derivatives checked

• MaziyarPanahi/MeliodasM7-7B-GGUF: Same weights, converted

On the Hub · Downloads on Model Pulse