title: Distilbert Base Uncased
emoji: 🐠
colorFrom: yellow
colorTo: gray
sdk: streamlit
sdk_version: 1.41.1
app_file: app.py
pinned: true
license: mit
thumbnail: >-
/static-proxy?url=https%3A%2F%2Fcdn-uploads.huggingface.co%2Fproduction%2Fuploads%2F64fbe312dcc5ce730e763dc6%2FIi2w5OVuSGhtPwC76W83c.png%3C%2Fspan%3E
To deploy your project on Streamlit, you'll need to create two essential files: app.py and requirements.txt.
1. app.py
This Python script serves as the main entry point for your Streamlit application. It should include the necessary imports and define the application's layout and functionality. Here's an example based on your project:
import streamlit as st
from transformers import AutoTokenizer, AutoModelForSeq2SeqLM
import torch
# Load the tokenizer and models
tokenizer = AutoTokenizer.from_pretrained("cssupport/t5-small-awesome-text-to-sql")
original_model = AutoModelForSeq2SeqLM.from_pretrained("cssupport/t5-small-awesome-text-to-sql", torch_dtype=torch.bfloat16)
ft_model = AutoModelForSeq2SeqLM.from_pretrained("daljeetsingh/sql_ft_t5small_kag", torch_dtype=torch.bfloat16)
# Move models to GPU
device = 'cuda' if torch.cuda.is_available() else 'cpu'
original_model.to(device)
ft_model.to(device)
# Streamlit app layout
st.title("SQL Generation with T5 Models")
# Input text box
input_text = st.text_area("Enter your query:", height=150)
# Generate button
if st.button("Generate SQL"):
if input_text:
# Tokenize input
inputs = tokenizer(input_text, return_tensors='pt').to(device)
# Generate SQL queries
with torch.no_grad():
original_sql = tokenizer.decode(
original_model.generate(inputs["input_ids"], max_new_tokens=200)[0],
skip_special_tokens=True
)
ft_sql = tokenizer.decode(
ft_model.generate(inputs["input_ids"], max_new_tokens=200)[0],
skip_special_tokens=True
)
# Display results
st.subheader("Original Model Output")
st.write(original_sql)
st.subheader("Fine-Tuned Model Output")
st.write(ft_sql)
else:
st.warning("Please enter a query to generate SQL.")
2. requirements.txt
This file lists all the Python packages your application depends on. Streamlit will use this file to install the necessary packages during deployment. Here's an example:
streamlit
transformers
torch
Ensure that the versions of the packages are compatible with each other and with your code. You can specify exact versions if needed, for example:
streamlit==1.15.2
transformers==4.11.3
torch==1.10.0
To generate a requirements.txt file with the exact versions of the packages installed in your environment, you can use the following command:
pip freeze > requirements.txt
This command will list all installed packages and their versions, which you can then include in your requirements.txt file.
For more information on creating and deploying Streamlit apps, refer to the official Streamlit documentation:
By setting up these files correctly, you can deploy your SQL generation application on Streamlit, allowing users to input queries and receive generated SQL statements from both the original and fine-tuned models.