CompressingVLM/qwen3-vl-2b-boundingdocs-ft-kd-bnb-nf4
Source model: /kaggle/working/dce_checkpoints/penten/model_kd_phase2
Quantization: bitsandbytes 4-bit.
Model kind: penten_ang Qwen3-VL-2B distilled KD phase1 plus bbox phase2
This repository stores model artifacts saved after loading with BitsAndBytesConfig.
The accompanying notebooks evaluate with the same branch metrics used before quantization:
ANLS, spatial precision/recall/F1, ANLS * Spatial F1, and bbox coverage. For the Qwen distilled model they also report value exact/contains match.
