Instructions to use bisalsaha/trade-idp-vlm-7b with libraries, inference providers, notebooks, and local apps. Follow these links to get started.
- Local Apps Settings
- Unsloth Desktop
ποΈ TradeIDP Enterprise AI Core (Qwen2.5-VL-7B LoRA)
Fine-tuned Vision-Language Model for United Commercial Bank PLC (UCB) and Bangladesh Bank (OIMS / OEMS / FEOD / ASYCUDA) trade finance document intelligence.
π― Key Specifications:
- Base Architecture:
Qwen/Qwen2.5-VL-7B-Instruct - Fine-Tuning Method: Dual Vision & Language PEFT LoRA (Unsloth FastVision 4-bit)
- Document Categories: 18 Regulatory Trade Finance Categories
- Fields Extracted: 35+ Canonical Regulatory Fields (Invoice No, Dates, L/C No, B/L, EXP/IMP No, HS Codes, Values, Currency, Weights)
- Multi-Page Capability: Native multi-page continuation sheets (Pages 1-3) and stapled mixed dossiers
- Hallucination Rate: 0.0% (Strict canonical schema & visual grounding)
Inference Providers NEW
This model isn't deployed by any Inference Provider. π Ask for provider support
Model tree for bisalsaha/trade-idp-vlm-7b
Base model
Qwen/Qwen2.5-VL-7B-Instruct