How Snowpark Container Services and W&B Accelerate Your AI Development
We're thrilled to announce that we're a launch partner for Snowflake's new Snowpark Container Services offering. Here's what you need to know:
Created on June 27|Last edited on June 29
Comment
Introduction
Deep learning and generative AI teams are looking to take advantage of their data assets to create enterprise-grade machine learning products. One common technology enterprises rely on to store, analyze, and build data-focused applications is Snowflake’s Data Cloud—a fully managed, multi-cluster shared data architecture that leverages the near-infinite scalability of the cloud.
At Snowflake Summit 2023, Snowflake announced the private preview of Snowpark Container Services, an extension of the Data Cloud to securely develop and deploy containerized machine learning and AI applications. Weights & Biases is proud to be a launch partner of Snowpark Containers, enabling ML engineers and data scientists to seamlessly and securely leverage the W&B platform entirely within your Snowflake account.
What is Weights & Biases?
Weights & Biases helps ML teams build better models faster and improve developer productivity across the entire ML Workflow. With just a few lines of code in your notebook, you can instantly debug, compare, and reproduce your models—including architecture, hyperparameters, git commits, model weights, GPU usage, datasets, and predictions—all while collaborating with your teammates and using the platform as your ML system of record.
W&B is trusted by more than 500,000 ML practitioners, including some of the most innovative companies and research organizations in the world developing generative AI and Large Language Models like OpenAI, Cohere, and MidJourney. To try it for free, sign up at Weights & Biases.
What is Snowpark Container Services?
Snowpark Container Services, currently available in private preview, is an extension of Snowflake’s processing engine that provides developers with the flexibility to deploy container images in Snowflake-managed infrastructure. Customer-provided container images can include code in any programming language and be deployed on compute pools with configurable hardware options, including GPUs, with the same ease of use, scalability, and consistent governance of the Snowflake Data Cloud.
With Snowpark Container Services, data science engineers can run more complex AI and ML jobs and logic within Snowflake, as well as any other components of an application, including its web-based interface.
With this additional flexibility, developers can further accelerate development and streamline their architecture by having a single platform with consistent data security and governance to run both the ETL processing and the models and components that use the prepared datasets.
W&B + Snowpark Container Services: Accelerate your AI development in Snowflake
Today, we’re announcing two capabilities that enable Snowflake users to leverage the suite of developer tools from W&B to accelerate the development of machine learning and generative AI algorithms.
- W&B Server availability on Snowpark Container Services: Users can now stand up the W&B server within their Snowflake account,
- W&B SDK in Snowflake Conda Channel: Users can now leverage the W&B SDK to add in W&B API calls into their Jupyter projects hosted in Snowpark.
W&B and Snowpark Container Services Demo: Train, tune, and debug an LLM to create SQL queries from natural language
To showcase the power of W&B and Snowflake together, watch our demo on how to build a conversational agent on top of a large language model (LLM) from OpenAI to generate SQL queries executed in Snowflake.
The demo leverages the recently-released W&B Prompts, which logs and visualizes the prompts and interactions between an LLM and the data in Snowflake, then compares the results of the prompts, including the SQL generated by the LLM.
A sample trace timeline in W&B Prompts
The W&B platform also supports organizations who will be training and tuning their own models, including by leveraging NVIDIA’s NeMo Framework, and organizations who will be fine-tuning foundational LLM models, including by leveraging products such as NVIDIA’s NeMo LLM cloud service.
How to try it out
W&B on Snowpark Container Services is currently in private preview. If you’d like to learn more, please contact W&B. If you’d like to get a feel of the W&B Prompt, check out a sample live dashboard here (you’ll need to sign up for a W&B account).
Add a comment
Tags: Articles
Iterate on AI agents and models faster. Try Weights & Biases today.