guendilf/nim

Open source implementation for NIM: Neural Input Modeling

NIM

Open source implementation for NIM: Neural Input Modeling

Usage

1. Install NIM: pip install git+https://github.com/cenwangumass/nim.git
2. In the examples directory, create a synthetic dataset with 1,000 length 100 sequences and values are sampled from i.i.d. Gamma: python create_data.py -n 1000 -t 10 data.npy
3. Train the neural network: python train.py nvm data.npy --transform transform:log_nvm. For more options, run python train.py --help
4. Generate new samples and save in samples.npy: python generate.py nvm.json samples.npy -n 1000 -t 10 --transform=transform:exp