REL-DCPS-NLP

Repository for DCPS session on using NLP to improve analysis of goals

0. Download this repository using git

If neccesary, install git here. Then, in the folder you'd like to store the code and models, run the command git clone https://github.com/mathematica-pub/REL-DCPS-NLP-public.git using git bash (open git bash by right-clicking empty space in windows explorer and clicking "Open Git Bash here" in the menu that appears).

1. Download the GGUF Meta-Llama-3-8B quantized model from here.

On your local Windows machine, you should probably stick with something like the Q3_K_M model here.

NOTE: Files bigger than 4.30 GB are hard to run on Windows

2. Install anaconda

3. Using powershell, create a conda environment using the provided environment.yml file (this may take >15 minutes):

$ conda env create -f environment.yml
$ conda activate DCPS-NLP

4. In anaconda navigator, select the DCPS-NLP environment and open Jupyter

5. Use a Jupyter notebook to explore the commands in the .py files in this repository.