Cache for LLM calls

Setting up environment

conda create -n cache python=3.8
conda activate cache
conda install pip
pip install torch torchvision torchaudio --extra-index-url https://download.pytorch.org/whl/cu116
pip install -r requirements.txt

Usage

Export the arguments you want to change with respect to default values, following scripts/run.sh. For instance,

export TASK_NAME=isear
export BUDGET=1500

We use Neptune for logging. We recommend setting file scripts/cluster.sh with the following arguments:

export DATA_PATH=<YOUR_DATA_PATH>
export PROJECT_NAME_NEPTUNE=<YOUR_NEPTUNE_PROJECT_NAME>
export API_TOKEN_NEPTUNE=<YOUR_NEPTUNE_API_TOKEN>

Then, run the code:

bash scripts/run.sh