New in October: Find community with The DSC

October was a big month for the ML industry, with more momentum than ever behind spooky-good models and novel applications. And it was a big month for Baseten, as we launched a ton of features plus The DSC, a new community for data scientists led by our very own Jesse Mostipak.

What we’re reading, writing, and watching

Work with the world’s best models

Powerful open-source models like Stable Diffusion and Whisper have driven a new wave of interest in AI and ML, with builders around the internet creating exciting demos and promising products. You can use these and dozens of other open-source models in Baseten to build ML-powered apps.

Whisper model page

On our public model explore page, you can learn how to deploy and use models with revitalized READMEs that take you from deployment to invocation in seconds. New READMEs include invocation instructions, sample input and output, configuration notes, and use case ideas.

And if you’re interested in more advanced use cases for powerful models like these, reach out to Abu to learn more about our upcoming projects in this space.

Deploy MLflow models on Baseten

MLflow is an incredibly popular and powerful open source platform for training and managing models. As of Truss v0.1.4, you can package an MLflow model as a Truss and deploy it to Baseten. 

import truss

model = mlflow.pyfunc.load_model(MODEL_URI)
tr = truss.mk_truss(model, target_directory="./mlflow_truss")

Briefly, this integration works by wrapping your MLflow model in a pyfunc object, which is used in the Truss’ predict function. You can learn more in the Truss docs. Or, try it for yourself with this Google Colab notebook.

Community news

We’re so excited to share that we’ve launched The Data Science Community, a Discourse-based forum designed to be a warm, welcoming gathering place for data scientists and friends. You can sign up today by going to https://community.baseten.co/.

We’ve built The Data Science Community to be a place to share your expertise, ask for advice, and connect with other data friends from around the globe. There are categories such as Data Science and Machine Learning, MLOps, and Tooling to get conversations started within these topics, as well as a General category for conversations that don’t quite fit into any of the above categories. 

To kick things off we’ve started two reading groups, and we invite you to join both of them! Our bi-weekly reading group focuses on an open-source paper, blog post, video, or podcast, while our monthly reading group selects a longer-form book that you may need to purchase or borrow from the library.

Our bi-weekly group is currently reading and discussing Operationalizing Machine Learning: An Interview Study by Shankar, et. al., and our November book club will be reading The Data Detective by Tim Harford (if you’re outside North America you can find the book under the title How to Make the World Add Up). All of our discussions take place asynchronously within our Community forums, so you never have to worry about time zone math!

However you choose to participate in The Data Science Community, we’re excited to have you join us 💚

Inside Baseten

This month we welcomed Zack, a software engineer, who joined us from Amazon Prime Video. And colleagues old and new traveled to Mexico City at the end of October for an offsite full of food, fun, friends, and foundational work on our product and operations. We’d love to have you along at the next one, and we are actively hiring for a Lead ML Engineer and an Account Executive

This week, keep an eye on our Twitter … we’ve been building so much we’ll be publishing a new changelog daily for the rest of the week. And even more models, applications, and community stories are coming in November.

Thanks all,

The team at Baseten