MLReef is a holistic and open source platform with an ever expanding set of features to meet your ML requirements.
@data_processor(
name="Resnet50",
author="MLReef",
command="resnet50",
type="ALGORITHM",
description="CNN Model resnet50",
visibility="PUBLIC",
input_type="IMAGE",
output_type="MODEL"
)
@parameter(name='input-path', type='str', required=True, defaultValue='train', description="input path")
@parameter(name='output-path', type='str', required=True, defaultValue='output', description="output path")
@parameter(name='height', type='int', required=True, defaultValue=224, description="height")
@parameter(name='width', type='int', required=True, defaultValue=224, description="width")
@parameter(name='epochs', type='int', required=True, defaultValue=5, description="epochs")
@parameter(name='channels', type='int', required=False, defaultValue=3, description="channels")
@parameter(name='use-pretrained', type='str', required=False, defaultValue='True', description="pretrained")
@parameter(name='class-mode', type='str', required=True,defaultValue='sparse',description="classmode")
@parameter(name='batch-size', type='int', required=False, defaultValue=32, description="batch-size")
@parameter(name='validation-split', type='float', required=False, defaultValue=.2, description="validationsplit")
@parameter(name='learning-rate', type='float', required=False, defaultValue=.0001, description="learning rate")
@parameter(name='loss', type='str', required=False, defaultValue="sparse_categorical_crossentropy",
description="loss")
def init_params():
pass
Create or upload your scripts and add parameter decorators before publishing your code. The publishing will create a docker image so that your code is directly accessible within an interactive UI.

Each published module will be available as working code for you, your team or the entire MLReef community. Change its parameters and execute them in the built-in pipelines.

Track all jobs and experiments with GIT to reproduce data, code, hyperparameter values and environment settings used. Compare metrics and keep track of your progress.
Supported and powered by
View source code & contribute
A structured and containerized approach to Machine Learning development
Fast-track your ML work with a contextually structured development environment using built-in piplines and containerized scripts. Use CI/CD principles to efficiently operationalize your ML workflow.


Collaborative, modular and reusable ML content
Access community repositories with data sets, models, data operations and visualizations and reuse them to jumpstart your ML projects!
A full set of tools for your ML work!
Instead of disconnected toolchains, use MLReef to run your pipelines, process your data, version and track your experiments and model, manage your teams and members, ... and much more..

for everyone
MLReef supports open source for free. Get involved to perfect your craft and be part of something big.

for individuals
Use MLReef for your own ML projects, from experimental to hosting your life´s work.

for teams
Business of all sizes use MLReef to securely and efficiently develop their ML projects.
Keep informed about news and development progress in our newsletter!
No worries, we don't spam - just the most relevant news to keep up to date.