Uploaded finetuned model
- Developed by: puvith
- License: apache-2.0
- Finetuned from model : unsloth/qwen3-vl-8b-instruct-unsloth-bnb-4bit
This qwen3_vl model was trained 2x faster with Unsloth and Huggingface's TRL library.
Qwen3-8B Instruct POD Analyzer v3
Fine-tuned Qwen3-VL-8B-Instruct model for Proof of Delivery (POD) document analysis and information extraction.
- Base Model:
unsloth/Qwen3-VL-8B-Instruct-unsloth-bnb-4bit - Fine-tuned Model:
puvith/qwen3-8b-instruct-pod-analyzer_v3 - Task: POD document understanding
- Modality: Vision + Language
Intended Use
The model is designed to analyze POD images and extract visually supported information, including:
- Printed and handwritten information
- Signatures and stamps
- Clearly visible fields
- Partially visible or unclear fields
- Missing information
The model is instructed to avoid inventing or inferring information that is not visible in the document.
Training
The model was fine-tuned on a POD-specific annotated dataset using supervised fine-tuning with the Unsloth 4-bit Qwen3-VL-Instruct model.
Limitations
Performance may decrease with low-resolution, blurred, rotated, or obstructed POD images. Outputs should be validated before production use.
Base Model
unsloth/Qwen3-VL-8B-Instruct-unsloth-bnb-4bit
Model
puvith/qwen3-8b-instruct-pod-analyzer_v3
