Use GitHub-hosted demo video, replace large mp4 with 720p

Author: Yuxin

README.md

@@ -23,11 +23,9 @@ Code and data for [ICML 2026] **CoDA-Bench: Can Code Agents Handle Data-Intensiv
 CoDA-Bench (Code and Data-intensive Benchmark) is a benchmark for evaluating AI agents on data-intensive analytical tasks.
 Given a *natural language question* and access to a *Linux sandbox* containing hundreds of data files, an agent must discover relevant data, write code, and produce the correct answer.
 
-<p align="center">
-  <video src="https://github.com/ruc-datalab/CoDA-Bench/raw/main/assets/demo_video.mp4" controls width="80%"></video>
-</p>
+https://github.com/user-attachments/assets/5a34d86f-d1d6-49f9-bddb-7e023d7e9347
 
-> ▶️ Demo video. If it does not play inline, [watch it here](https://github.com/ruc-datalab/CoDA-Bench/raw/main/assets/demo_video.mp4).
+> ▶️ Demo video. If it does not play inline, [download/watch it here](./assets/demo_video_720p.mp4).
 
 Unlike existing benchmarks that provide oracle data directly, CoDA-Bench requires agents to:
 - 🔍 Discover relevant data among hundreds of semantically similar files