RESMPDEV/Qwen2-Wukong-5B-Instruct-Sliced

Partial fine-tune vs Qwen/Qwen2-7B-Instruct. Partial change: 36% of the weights changed (attention, mlp, norms), in 9 of 20 layers.

Tensors changed: Embeddings 0/1, Norms 18/41, MLP 27/60, Attention 63/140, lm_head 0/1

On the Hub · Downloads on Model Pulse