issai/Qolda-AVL-5B
Full fine-tune (fewer layers) vs Qwen/Qwen3-VL-4B-Thinking. Qwen/Qwen3-VL-4B-Thinking matches part of issai/Qolda-AVL-5B (713 of its 1224 tensors; the rest of issai/Qolda-AVL-5B has no counterpart). On the shared part: Fine-tune: 93% of the weights changed, across 36 of 36 layers.
Tensors changed: Embeddings 1/4, Norms 147/249, MLP 124/220, Attention 144/240
1 derivatives checked
- issai/Qolda-AVL-5B-GGUF: Same weights, converted (fewer layers)