unicorn-team/Unicorn-VL-R3

Partial fine-tune vs Qwen/Qwen3-VL-8B-Thinking. Partial change: 79% of the weights changed (attention, mlp, norms), in 36 of 36 layers.

Tensors changed: Embeddings 0/4, Norms 61/261, MLP 108/232, Attention 144/252, lm_head 0/1

2 derivatives checked

• mradermacher/Unicorn-VL-R3-i1-GGUF: Quantization (fewer layers)
• mradermacher/Unicorn-VL-R3-GGUF: Same weights, converted (fewer layers)

On the Hub · Downloads on Model Pulse