Welcome to Optimal-Demo-Selection-ICL Discussions!

[SatvikPraveen]

👋 Welcome to the Optimal-Demo-Selection-ICL Discussions
Hello everyone!

This space is dedicated to everything related to Optimal-Demo-Selection-ICL — a research-focused project exploring demonstration selection strategies for In-Context Learning (ICL) using Large Language Models (LLMs). Whether you're here to dive into the methods, replicate results, share feedback, or ask for help — you’re in the right place!

💡 What you can do here:

• ❓ Ask Questions about implementation, datasets, or paper-based methods
• 🐛 Report Issues or inconsistencies in code logic (use Issues tab for bugs!)
• 📈 Suggest Improvements or new experimental ideas
• 📚 Share Papers or resources relevant to ICL and LLMs
• 🧪 Showcase Results if you extended or reused the code
• 🌐 Collaborate with fellow contributors or researchers

🚀 About Optimal-Demo-Selection-ICL
This project implements and benchmarks optimal example selection techniques such as:

• IDS (Iterative Demonstration Selection)
• RDES (Reinforcement Demonstration Selection)
• Se² (Sequential Selection with Beam Search)
• TopK+ConE
• Influence-Based Selection

...across datasets like AG News, SST-5, and CommonsenseQA, using models such as LLaMA-3.2-3B, Gemma-2B, and GPT-2.

📦 GitHub Repo: Optimal-Demo-Selection-ICL

Feel free to start a discussion thread below or explore the existing ones.

Let’s push the frontier of ICL research together! 💬🧠✨

– The Optimal-Demo-Selection-ICL Team