ICLR 2025 Paper Search

This Flask app allows users to search for ICLR 2025 paper submissions using semantic search and BM25 ranking.

Deployed at https://iclr-search.jkapoor.me

Prerequisites

• Python 3.7 or higher
• Vercel CLI
• OpenAI API key

Local Development

1. clone this repository:

   git clone <repository-url>
   cd <repository-name>
   

2. install the required packages:

   pip install -r requirements.txt
   

3. set up your openai api key as an environment variable:

   export OPENAI_API_KEY=your_api_key_here
   

4. run the flask app locally:

   python api/index.py
   

5. open your browser and go to http://localhost:5000 to use the app.

Deploying to Vercel

1. install the vercel cli:

   npm i -g vercel
   

2. login to vercel:

   vercel login
   

3. deploy the app:

   vercel
   

4. set the openai api key as an environment variable on vercel:

   vercel env add OPENAI_API_KEY
   

   enter your openai api key when prompted.

5. redeploy the app to use the new environment variable:

   vercel --prod
   

Your app should now be deployed and accessible via the Vercel URL provided.

Files

• api/index.py: main flask application
• vercel.json: vercel configuration file
• requirements.txt: python dependencies
• iclr_2025_submissions.json: json file containing paper submissions (precomputed in this rep)
• embedding_array_fp16.npy: numpy array of paper embeddings (precomputed in this rep)

Make sure to include the iclr_2025_submissions.json and embedding_array_fp16.npy files in your repository or upload them to Vercel separately.

Note

Ensure that your OpenAI API key has sufficient credits and permissions to create embeddings using the "text-embedding-3-small" model.