# PITT

**Repository Path**: myendless/pitt

## Basic Information

- **Project Name**: PITT
- **Description**: Reproduction and further exploration work of physic informed token transformer
- **Primary Language**: Unknown
- **License**: Not specified
- **Default Branch**: fusion
- **Homepage**: None
- **GVP Project**: No

## Statistics

- **Stars**: 0
- **Forks**: 0
- **Created**: 2023-08-26
- **Last Updated**: 2023-08-31

## Categories & Tags

**Categories**: Uncategorized

**Tags**: None

## README

# PITT
Reproduction and further exploration work of physic informed token transformer

# 2023/8/21 milestone content:
+ standard PITT content
+ fusion training
+ loss comparison of fusion and other models on three 1D datasets

P.S. Directory /pde_data including 1D data files are not uploaded because of their huge size.

# how to evaluate fusion and other trained models?
1. download 1D data files.
2. run train_pitt.py to get heat, burgers and KdV models, configs are processed automatically.
3. run train_pitt_1d_fusion.py to get fusion model.
   + trained models are in /1D_results for fusion, heat, burgers and KdV(part) already.
4. run compare_models_1D_num_sample.py to evaluate models