Repository for DCPS session on using NLP to improve analysis of goals
- 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).
- 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
-
Install anaconda
-
Using powershell, create a
conda
environment using the providedenvironment.yml
file (this may take >15 minutes):
$ conda env create -f environment.yml
$ conda activate DCPS-NLP
-
In anaconda navigator, select the DCPS-NLP environment and open Jupyter
-
Use a Jupyter notebook to explore the commands in the
.py
files in this repository.