Skip to content

Latest commit

 

History

History
25 lines (22 loc) · 912 Bytes

File metadata and controls

25 lines (22 loc) · 912 Bytes
title Streanlit GEMMA 2B & Deepseek Coder 1.3B
emoji 📉
colorFrom gray
colorTo gray
sdk streamlit
sdk_version 1.31.1
app_file app.py
run_code streamlit run app.py
pinned false
license Apache 2.0

Check out the configuration reference at https://huggingface.co/docs/hub/spaces-config-reference

Speak2SQl

This web app is built to convert the natural language to sql queries. It is powered by 2 state of the art Transformer model,

  • GEMMA 2B
  • Deepseek Coder 1.3B

Both the models are fine tuned over 3000 rows and have been tested againest 100 seperate rows. The evaluation metrics used for evaluating these models are ~

  • BLEU SCORE
  • SCAR BLEU SCORE
  • EM RATIO

Here BLEU SCORE considers the ngram probability and checks that if the predicted text matches with the actual text over n-gram, whereas EM Ratio checks that the predicted query exactly matches with the actual query or not.