jakeatx/slimder-qwen38-reap384-depth32-mb2-3-4-5

Partial fine-tune vs jakeatx/slimder-qwen38-reap384-depth40-mb4-5. Partial change: 41% of the weights changed (attention, mlp, norms), in 24 of 32 layers.

Tensors changed: lm_head 0/1, Embeddings 0/132, Norms 90/124, other 0/5, Attention 246/328, MLP 240/320

1 derivatives checked

On the Hub · Downloads on Model Pulse