Spaces:
Running
Apply for a GPU community grant: Personal project
Description of the app: This public open-source demo extracts structured JSON from Russian-language procurement and corporate contracts using the zieglerd/RussianConcsExt LoRA adapter for Qwen2.5-32B-Instruct. It is intended for researchers, document-processing practitioners, and users evaluating Russian legal information extraction. The complete Gradio/ZeroGPU application is already included as app.py and requirements.txt; the Space currently uses a static landing page because this non-PRO account cannot create a Gradio CPU Space.
Justification: The 32B base model requires GPU inference. The demo uses 4-bit NF4 quantization and fits the 48 GB ZeroGPU large tier. A community grant would make this open model publicly testable without paid dedicated hardware. Please grant ZeroGPU access and allow this Space to switch from Static to Gradio; GRADIO_README.md contains the ready configuration.