sunkaianna / BiLLP

Overview

• Define tasks in tasks/
• Collect data & run experiments via generation_rec_agents.py
• Results will be saved in trajs_agent/

Data & Prompts

• Data to generate training data and run experiments in env/.
• Download data from steam_test.npy, steam_train.npy, test_distance_mat.pickle, train_distance_mat.pickle to /env/steam/

Setup

Set up OpenAI API key and store in environment variable (see here)

export OPENAI_API_KEY=<YOUR_KEY>

Create virtual env, for example with conda

conda create -n BiLLP python=3.9
conda activate BiLLP

Install dependencies

pip install -r requirements.txt

Run Demo

Grounding Embedding Generation

Generate embedding via /env/steam/llama_generate_embedding.ipynb

Example:

train and test

source run_steam.sh
source run_steam_test.sh

source run_amazon.sh
source run_amazon_test.sh

References

Our code is based on FireAct