Assets

... optimizing your manufacturing processes with modular digital twins composed of expert assets...

icon

Avalanche-MS

CL framework

Microservice

Provided by SZTAKI - Institute for Computer Science and Control 8 months, 1 week ago (last modified 7 months, 3 weeks ago); viewed 244 times and bound 1 time

Description:

Avalanche is an open-source end-to-end Continual Learning library based on PyTorch. Its goal is to support fast prototyping, training and evaluation of continual learning algorithms. Its initial focus was on continual supervised learning for vision tasks. It focuses on reproducibility, scalability and code efficiency (e.g., requiring less code, allowing faster iteration and reducing errors).

Input parameters:

JUPYTERLAB_PASSWORD: Password for JupyterLab interface

DATA_PATH: The data path specifies the location of the avalanche container, where data is stored, and it should be synchronized with Rclone MS

REGISTRY_PATH: The registry path indicates the model location and should be synchronized with Rclone MS

Default port(s):

8888: The JupyterLab interface.

---

https://avalanche.continualai.org/

https://github.com/ContinualAI/avalanche

Classification: Other

Type: Other

Expected User's Input Parameters
JUPYTERLAB_PASSWORD

Password for JupyterLab interface

  • Type: text
  • Mandatory: No
DATA_PATH

The data path specifies the location of the avalanche container, where data is stored, and it should be synchronized with Rclone MS

  • Type: text
  • Mandatory: Yes
REGISTRY_PATH

The registry path indicates the model location and should be synchronized with rclone MS.

  • Type: text
  • Mandatory: Yes
Container Deployment Information
  • Format: docker-compose
  • Data:
    01. services:
    02.   avalanche:
    03.     environment:
    04.     - JUPYTERLAB_PORT=8888
    05.     - JUPYTERLAB_PASSWORD=${JUPYTERLAB_PASSWORD}
    06.     image: dbs-container-repo.emgora.eu/avalanche:0.4.0
    07.     ports:
    08.     - mode: host
    09.       protocol: TCP
    10.       published: '8888'
    11.       target: '8888'
    12.     privileged: true
    13.     volumes:
    14.     - bind:
    15.         propagation: rshared
    16.       source: ${DATA_PATH}
    17.       target: /data
    18.       type: bind
    19.     - bind:
    20.         propagation: rshared
    21.       source: ${REGISTRY_PATH}
    22.       target: /model
    23.       type: bind
    24. version: '3.9'
    25. 
  • GUI Microservice: Yes
  • Workload type: service
  • Opened ports: 8888