This repo is for playing with llama-2 (Huggingface version).
You can browse data or self-annotate data via the link below: https://polito-demo-deployment.streamlit.app/
- LLaMA2 Demos on SST (Inference and Fine-tuning).
- LLaMA2 application on politosphere dataset.
- The usage of Prompt Template Tool .
- (Still Adding....)
Since models need to be run on the IC Cluster, I cannot perform local Huggingface/Wandb logins in the terminal/image. Therefore, I have outlined in the documentation (project overview -> RUNAI Command) how to pass environment variables using RUNAI. You can also see how to read environment variables with Python codes in files under this repo. However, in some code files, for the sake of convenience during testing, I've exposed my token. Please exercise extreme caution before using it.
Yifei Song (yifei.song@epfl.ch)