Sign inSign up

mfekadu/nimbus-elastic-tf-embed-worker

By mfekadu

•Updated about 6 years ago

Image
0

887

mfekadu/nimbus-elastic-tf-embed-worker repository overview

⁠nimbus

⁠Demo⁠

⁠Documentation⁠

⁠TL;DR;

⁠if ONLY CPU with Ubuntu 18.04 on Google Cloud Platform

NOTE: 🙏 Please change the passwords in the .env file

git clone https://github.com/mfekadu/nimbus \
    && cd nimbus \
    && ./ubuntu_setup.sh \
    && cp .env_sample .env \
    && ./deploy.sh
⁠elif GPU with Ubuntu 18.04 on Google Cloud Platform
git clone https://github.com/mfekadu/nimbus \
    && cd nimbus \
    && ./gpu_setup.sh \
    && ./ubuntu_setup.sh \
    && cp .env_sample .env \
    && ./deploy.sh
⁠Else Keep Reading

⁠Hardware Requirements

click to expand

The following requirements are based on Rasa-X documentation⁠

vCPUs (virtual CPUs on Google Cloud Platform or equivalent number of cores)

  • Minimum: 2 vCPUs
  • Recommended: 2-6 vCPUs

RAM

  • Minimum: 4 GB RAM
  • Recommended: 8 GB RAM

Disk Space

  • Recommended: 100 GB disk space available

Open These Ports (if running on Google Cloud Platform)

  • 22 (SSH)
  • 80 (HTTP)
  • 443 (HTTPS)
  • 5002 (rasa-x)
  • 5065 (rasa-production)
  • 5075 (rasa-worker)
  • 5055 (app) (aka rasa-actions service)
  • 5432 (db) (postgres)
  • 5672 (rabbit)
  • 8000 (duckling)
  • 6379 (redis)
  • 9200 (elasticsearch-service)
  • 9300 (elasticsearch-service)
  • 9010 (nimbus-elastic-tf-embed-worker)

⁠Software Requirements

click to expand

Python 3.6.8 or newer

Docker 19.03.8 or newer

docker-compose 1.25.5 or newer

$ python3 --version

Python 3.6.8

$ docker -v

Docker version 19.03.8, build afacb8b

$ docker-compose -v

docker-compose version 1.25.5, build 8a1c60f6

⁠Getting Started

⁠1. Install Docker Engine⁠

It works with Windows⁠ / macOS⁠ / Linux (Ubuntu preferred).⁠.

Carl Boettiger⁠ suggests that Docker "could have promising implications for reproducible research in scientific communities"⁠ despite challenges like adoption by the scientific community. (arxiv⁠).

⁠2. Install Docker-Compose⁠

NOTE: The previous step may have included the docker-compose binary depending on your operating system.

⁠3. Clone this Repository
click to expand 
git clone https://github.com/mfekadu/nimbus
⁠4. Create the .env file with the following secrets

The .env file is read by docker-compose⁠ (from current working directory) to string-replace the variables in the docker-compose.yml⁠ file.

See the .env_sample⁠

TODOs:

  • make actions image a variable and put in .env_sample
⁠4.5. Resources for Docker Desktop on (macOS / Windows)
click to expand

Sorry that this system is resource intensive. There may be room for optimization!

Docker Desktop Resources

⁠5. Run all the services

The following command will download/cache/install/build -- do everything to set up the infrastructure and run the program.

Run the following command in the root of the directory for this codebase:

docker-compose up --detach

NOTE: run with sudo if need be on Linux/macOS

NOTE: Grab a coffee or some snacks. The various downloads will take some time, but docker will cache each stage of the build, so things will be faster next time.

⁠6. Download the latest chatbot model⁠

The latest model has the format 1-latest-YYYYMMDD-SSSSSS.tar.gz

https://drive.google.com/drive/folders/1nUG5g63mPg3GN1CH1I6Z3uMUsMEQlOqS
⁠7. Chat

A custom chatbot UI can be found at

http://localhost

The chatbot admin page can be accessed after logging in with RASA_X_PASSWORD found in .env⁠:

http://localhost:5002

Tag summary

Content type

Image

Digest

Size

1.4 GB

Last updated

about 6 years ago

docker pull mfekadu/nimbus-elastic-tf-embed-worker