xlangai/Jedi-3B-1080p
Partial fine-tune vs Qwen/Qwen2.5-VL-3B-Instruct. Partial change: 83% of the weights changed (attention, embeddings, mlp, norms), in 36 of 36 layers.
Tensors changed: Embeddings 1/2, Norms 73/138, MLP 112/304, Attention 252/380
3 derivatives checked
- bartowski/xlangai_Jedi-3B-1080p-GGUF: Same weights, converted (fewer layers)
- mradermacher/Jedi-3B-1080p-GGUF: Same weights, converted (fewer layers)
- lmstudio-community/Jedi-3B-1080p-GGUF: Quantization (fewer layers)