cagataydev/loopl-vl-2b
Partial fine-tune vs Qwen/Qwen3-VL-2B-Instruct. Partial change: 81% of the weights changed (attention, embeddings, mlp, norms), in 28 of 28 layers.
Tensors changed: Embeddings 1/4, Norms 86/217, MLP 84/196, Attention 112/208
2 derivatives checked
- mradermacher/loopl-vl-2b-i1-GGUF: Quantization (fewer layers)
- mradermacher/loopl-vl-2b-GGUF: Same weights, converted (fewer layers)