Spaces:
Running on Zero
Running on Zero
Add app.py, requirements.txt, schema.sql, README.md to serve the fhirsql model via ZeroGPU
3434a32 verified |
Download README.md from adelelsayed1991/fhirsql-reasoning-sql-adapters: direct link, hf CLI and curl.
- Browser
- Download file 731 Bytes
-
https://huggingface.co/spaces/adelelsayed1991/fhirsql-reasoning-sql-adapters/resolve/main/README.md
- Command line
-
hf download hf://spaces/adelelsayed1991/fhirsql-reasoning-sql-adapters/README.md
-
curl -L -o README.md https://huggingface.co/spaces/adelelsayed1991/fhirsql-reasoning-sql-adapters/resolve/main/README.md
731 Bytes
A newer version of the Gradio SDK is available: 6.29.1
metadata
title: fhirsql-reasoning-sql-adapters
emoji: 🏥
colorFrom: blue
colorTo: green
sdk: gradio
app_file: app.py
pinned: false
fhirsql-reasoning-sql
Translates a natural-language hospital question into a single DuckDB SQL
statement, using a DoRA/LoRA fine-tune of Qwen2.5-Coder-14B-Instruct
(adapters: adelelsayed1991/fhirsql-reasoning-sql-adapters) against the
fhirsql project's FHIR-derived schema.
Runs on ZeroGPU. Returns the literal string UNANSWERABLE when the
question cannot be answered from the schema shown to the model.
Reproduces the exact system prompt and chat-template format used during
training (see the fhirsql project's sft_train.ipynb) so generations
here match the fine-tuning distribution.