Update readme.

README.md

@@ -1,9 +1,46 @@
-# Pipeline Parallelism Emulation and Visualization
+---
+title: Pipeline Parallelism Schedule Visualizer
+emoji: 📊
+colorFrom: indigo
+colorTo: blue
+sdk: docker
+sdk_version: latest
+app_file: app.py
+pinned: false
+---
 
-This project provides tools for emulating and visualizing pipeline parallelism strategies used in large language model training.
+# Pipeline Parallelism Schedule Visualizer
+
+An interactive visualization tool for exploring different pipeline parallelism scheduling strategies in large language models.
+
+## Features
+
+- Visualize multiple scheduling strategies for pipeline parallelism
+- Adjust parameters like number of devices, stages, and batches
+- Compare execution timelines between different strategies
+- Explore operation timings and their effects on performance
+
+## Supported Strategies
+
+- 1F1B (One-Forward-One-Backward)
+- 1F1B with Interleaved Placement
+- 1F1B with Overlapped Operations
+- 1F1B with Interleaved Placement and Overlapped Operations
+- Zero-Bubble 1 Pipeline (ZB1P)
+- Dual Pipeline (DualPipe)
+
+## Usage
+
+Simply adjust the parameters and select the strategies you want to compare, then click "Generate Schedule" to visualize the results.
+
+## Deployment
+
+This app is deployed on Hugging Face Spaces using Dash.
 
 ## Overview
 
+This project provides tools for emulating and visualizing pipeline parallelism strategies used in large language model training.
+
 Pipeline parallelism is a technique used to train large models by partitioning the model across multiple devices and processing data in a pipelined fashion. This project allows you to:
 
 - Simulate different pipeline parallelism strategies (1F1B, Interleaved, Zero-Bubble, etc.)