Add Quickstart Notebook for Beginners (Fixes #3)

[AbhayP1604]

Description

I have added a QuickStart.ipynb notebook to help new users run the mini-VLA pipeline end-to-end without needing complex terminal commands.

Changes Made

• Created a step-by-step Jupyter Notebook covering:
  1. Environment Setup: Installs requirements directly.
  2. Data Collection: Runs the expert script (captured 10 episodes for demo).
  3. Training: Trains the Diffusion Policy (configured for 5-50 epochs).
  4. Testing: Runs the inference loop and saves videos.
  5. Visualization: Embeds and plays the resulting videos directly in the notebook.

Notes for Reviewers

• Warnings: I added code to suppress the gymnasium and metaworld version mismatch warnings to keep the output clean for users.
• Camera View: The data collection uses --camera-name corner (Side View), but the testing script currently defaults to Top View (since test.py does not seemingly accept a camera argument). I noted this behavior in the notebook.

Checklist

• Notebook runs successfully on CPU.
• Output videos are generated and displayed.
• Documentation added within the notebook to explain steps to beginners.

Fixes #3

[keivalya]

Lol no, sorry but this is not what I meant. Just pushing in CLIs wrapped inside tiny blocks is not gonna help anyone. Sure, it might work -- but quite useless -- you can't explain code from there. That's what makes python notebooks useful -- so that you can teach tiny code-blocks independently.

I appreciate your contribution, thanks for putting your time and efforts.

:D

[AbhayP1604]

Yeah I'm sorry but this is what I also thought like the notebook I made is basically rerun/walkthrough of readme.md kinda redundant so...Now my question is how a quickstart nb should really ? And btw I new to git/open-source space if could guide me a bit I would be very grateful.

Also I actually read all the blogs you posted about mini-Vla and a bit about CNNs - was a very good read.