You can start by cloning our repository and following the steps below.

1. Install the dependencies for our code using Conda. You may need to adjust the environment YAML file depending on your setup.

   conda env create -f environment.yml
   conda activate in-context-learning
   

2. Download model checkpoints and extract them in the current directory.

   wget https://github.com/dtsip/in-context-learning/releases/download/initial/models.zip
   unzip models.zip
   

3. [Optional] If you plan to train, populate conf/wandb.yaml with you wandb info.

That's it! You can now explore our pre-trained models or train your own. The key entry points are as follows (starting from src):

• The eval.ipynb notebook contains code to load our own pre-trained models, plot the pre-computed metrics, and evaluate them on new data.
• train.py takes as argument a configuration yaml from conf and trains the corresponding model. You can try python train.py --config conf/toy.yaml for a quick training run.

• Shivam Garg
• Dimitris Tsipras