janhq/Jan-v3.5-4B
Partial fine-tune vs janhq/Jan-v3-4B-base-instruct. Partial change: 82% of the weights changed (attention, mlp, norms), in 36 of 36 layers.
Tensors changed: lm_head 0/1, Embeddings 0/1, MLP 108/108, Norms 80/145, Attention 144/144
Partial fine-tune vs janhq/Jan-v3-4B-base-instruct. Partial change: 82% of the weights changed (attention, mlp, norms), in 36 of 36 layers.
Tensors changed: lm_head 0/1, Embeddings 0/1, MLP 108/108, Norms 80/145, Attention 144/144