Paper associated with project

Hypothetical_Tail_Embeddings.pdf

KGE-LLM

Environment setup

1. Ensure miniconda3 is installed
2. Source conda with source $YOUR_PATH/miniconda3/bin/activate
3. If you are creating the environment for the first time, navigate to the KGE-LLM root directory run conda env create -f environment.yml, if the environment is already created go to step 4
4. Run conda activate KGE

Hugging Face Access Token

1. To be able to run Llama models create a local .env file in the root directory of KGE-LLM
2. In the .env file create a variable HF_TOKEN = $YOUR_HF_TOKEN

Experiments

1. To run experiments run python -m experiments.openai_experiment
2. Ensure the conda environment is active
3. Ensure you have an OpenAI API Key entered in the code
4. Ensure you are in the root directory of the repository

Finetuning

1. To run finetuning run python -m finetune.finetuned-minilm