search

ludwig-ai / ludwig

Low-code framework for building custom LLMs, neural networks, and other AI models
http://ludwig.ai
Apache License 2.0
11.01k stars 1.18k forks source link

Sagemaker Support #614

Open bcordo opened 4 years ago

bcordo commented 4 years ago

What would be the best way to use this with AWS sagemaker? Would it even make sense to use it with Sagemaker, or are the Ludwig features duplicates?

I have found this: https://kenza-ai.github.io/sagify/ And resources such as this: https://sagemaker.readthedocs.io/en/stable/using_tf.html

I just wanted to get people's opinions first.

jenishah commented 4 years ago

This feature would be extremely useful.

w4nderlust commented 4 years ago

Taken a look at it. It seems pretty straightforward to implement the train and predict functions in sagify. They will probably end up 1/2 lines of code each calling LudwigModel.train and LudwigModel.predict`. This is a potential feature to add and if someone from the community wants to work on it, that would be great.

jenishah commented 4 years ago

I want to work on it, and I read the documentation for sagify. Are we supposed to push the final docker image to ECS?

brightsparc commented 3 years ago

@jenishah I have published a sample on how to build, train and deploy a Ludwig model with Amazon SageMaker if you want to check it out: https://github.com/aws-samples/amazon-sagemaker-ludwig-transformer