# CRT

**Repository Path**: sing_jay_lee/CRT

## Basic Information

- **Project Name**: CRT
- **Description**: No description available
- **Primary Language**: Unknown
- **License**: Not specified
- **Default Branch**: main
- **Homepage**: None
- **GVP Project**: No

## Statistics

- **Stars**: 0
- **Forks**: 0
- **Created**: 2020-11-26
- **Last Updated**: 2020-12-19

## Categories & Tags

**Categories**: Uncategorized

**Tags**: None

## README

# CRT: Causal Representation Transfer

Code for Supercharging Imbalanced Data Learning WithCausal Representation Transfer 

## Model

### Prerequisites

The algorithm is built with:

* Python (version 3.7 or higher)
* Numpy (version 1.16 or higher)
* PyTorch (version 1.3.1)

Clone the repository, e.g.:
```
git clone https://github.com/ZidiXiu/CRT.git
```


### Running the Toy Dataset

Here we present a toy synthetic dataset 

```
python train train_CRT_toy.py
```

When building the CRT framework, we referenced the following sources: 
* [MADE flow](https://github.com/kamenbliznashki/normalizing_flows/blob/master/maf.py)
* [GCL learning](https://github.com/takeshi-teshima/few-shot-domain-adaptation-by-causal-mechanism-transfer)