1. Screenshots
   

• ✅ Task-1: Create a repo and store your program on GitHub or Bitbucket. (https://github.com/CrypticMessenger/rapidfort)
• ✅ Task-2: Documentation. (README.md)
• ✅ Task-3: Add a simple UI to this web application. (frontend.py)
• ✅ Task-4: Dockerize the application. (Dockerfile)
• ✅ Task-5: Add a GitHub actions or equivalent pipeline to build a docker container. (.github/workflows/main.yml)
• ✅ Task-6: Create a bash script with instructions to run the container. (run.sh)
• ✅ Task-7: Create Kubernetes manifest files to host the web server. (pod.yaml)
• 🚀 BONUS: Dockerhub image is available for both arm64 and amd64 architectures(used github actions), making it more universal.
• 🚀 BONUS: deployed frontend on cloud using streamlit sharing. (https://rapidfort.streamlit.app/)
• 🚀 BONUS: Maintained high standards of code quality: using enviroment variables and saved coding time by using open-sourced templates.
• 🏆 PERSONAL ACHIEVEMENT: boosted testing time by downloading model once locally and then transfering it to container using, thus avoiding downloading it everytime the container is run: (expected time reduced from 50 minutes to 5-7 minutes)

docker cp ~/.cache/huggingface <container_id>:/root/.cache/

• 🏆 PERSONAL ACHIEVEMENT: never worked on docker, kubernetes or github actions before, so it was a great learning experience.

• Quick Links:

  • hosted frontend: https://rapidfort.streamlit.app/
  • dockerhub image: ankitsharma61016/gitagpt108:latest
  • Collab backend link: https://colab.research.google.com/drive/1XQ36LZJ_znkBaqkhHWivyYWbYpFdkxcG?usp=sharing
  • resume link: https://drive.google.com/file/d/1IQRxkd0824bWJVbNHF-_HWwhrofTR3hB/view?usp=drive_link

• Clone the repo using following command:

https://github.com/CrypticMessenger/LlamaYogi.git

• Run run.sh using following command:

./run.sh

if you get permission denied error, run the following command and try again:

chmod +x run.sh

• in background the script will do the following things:

  • Check if container with the specified name is already running. if yes, it will exit the script.
  • Check if the container exists but is stopped, it removes the container.
  • Pull the latest image from the repository
  • Run the Docker container with port mapping
  • Check if the container started successfully, and echo the result.
  • [CAUTION] you can change some parameters in run.sh but be very careful while doing so.

• to run frontend locally, run the following command, then go to localhost:8502 in your browser:

streamlit run frontend.py --server.port 8502

• hosted frontend on cloud using streamlit sharing, link: https://rapidfort.streamlit.app/

• [CAUTION] remember the llama2 model is very intensive and requires a lot of memory(13GB+). so, if you are running it on a local machine, it might crash. so, it is recommended to run it on a cloud instance, through google colab. the link to the colab notebook is given below, doing same thing as in backend.py but in cloud(without frontend): follow this article to generate HF_TOKEN(https://huggingface.co/blog/llama2)
  Collab link: https://colab.research.google.com/drive/1XQ36LZJ_znkBaqkhHWivyYWbYpFdkxcG?usp=sharing

• to create kubernetes cluster, run the following command:

kubectl create -f pod.yaml

• to check if the pod is running, run the following command:

kubectl get deployments

• to tear down the cluster, run the following command:

kubectl delete -f pod.yaml