protocolsyncllc/medgap-qwen-3.5-9b

Full fine-tune vs unsloth/Qwen3.5-9B-Base. Fine-tune: 96% of the weights changed, across 32 of 32 layers.

Tensors changed: lm_head 1/1, Embeddings 2/4, Norms 107/215, Attention 197/332, MLP 100/208

1 derivatives checked

• mradermacher/medgap-qwen-3.5-9b-GGUF: Same weights, converted

On the Hub · Downloads on Model Pulse