DiCoW JSALT25 Tutorial & Challenge

This repository contains materials for the JSALT 2025 Summer School Tutorial & Challenge on fine-tuning DiCoW in data- and compute-constrained environments.

All instructions, code, and steps are provided in the challenge.ipynb notebook.

---

🔗 Leaderboard

Submit your results to the official leaderboard: 👉 EMMA Leaderboard on Hugging Face

---

▶️ Watch the Tutorial Recording

You can watch the tutorial session here: 🎥 JSALT 2025 Tutorial Recording

---

🚀 Run Directly in Google Colab

To run the notebook in Colab, simply replace github.com with githubtocolab.com in the URL. For example:

https://github.com/Lakoc/JSALT_tutorial/blob/main/challenge.ipynb

becomes

https://githubtocolab.com/Lakoc/JSALT_tutorial/blob/main/challenge.ipynb

---

📦 About

This challenge focuses on:

• Efficient fine-tuning of DiCoW models
• Training with limited data & compute resources
• Overlapping speech transcription