Snowflake AI Data Cloud

Snowflake Cortex AI

Snowflake Cortex provides direct access to advanced large language models from top tech firms (Mistral, Reka, Meta, Google ) and its open model, Snowflake Arctic. Hosted by Snowflake, these models require no setup and operate within Snowflake's ecosystem, guaranteeing efficient performance, scalability, and governance.

COMPLETE

As the name suggests when complete is given a prompt, returns a response that completes the prompt. This function accepts either a single prompt or a conversation with multiple prompts and responses;

These models are

1. Meta (llama3-8b,llama3-70b),

2. snowflake-arctic,

3. Reka functions(reka-core,reka-flash),

4. Mistral(mistral-large, mixtral-8x7b.)

EMBED_TEXT_768/ EMBED_TEXT_1024

Given a piece of text, returns a vector embedding of 768 dimensions that represents that text.

2: Given a piece of text, returns a vector embedding of 1024 dimensions that represents that text.

EXTRACT_ANSWER

Given a question and unstructured data, returns the answer to the question if it can be found in the data.

SENTIMENT

Returns a sentiment score, from -1 to 1, representing the detected positive or negative sentiment of the given text.

SUMMARIZE

Returns a summary of the given text

TRANSLATE

Translates given text from any supported language to any other.

SNOWFLAKE CORTEX FUNCTIONS

Components of Snowflake AI

Cortex Analyst

Cortex Search

Cortex Guard

Cortex Guard as the name suggests seeks to make snowflake models safe and usable. This is done by employing filters to the LLMs inputs and outputs that are deemed harmful, violent and abusive etc, by flagging them as harmful content. 

Cortex Finetuning

Enables finetuning of LLMs using SQL functions or by Using Cortex Studio(Private Preview now)

Snowflake Machine Learning

Feature Store

The feature store to is a system that facilitates the creation, storage, management, and deployment of machine learning (ML) features. It ensures consistency in ML features used across training and inference stages. Additionally, it supports the automation of data refreshes, accommodating both batch and real-time streaming data, to maintain up-to-date and relevant ML models.

Help with the tracing of Features ,Datasets ,Models across end-to-end ML lifecycle

Notebooks provides a cell-by-cell development interface to explore and get insight from the data. No configuration is required, you can collaborate and integrate with Github, full RBAC-based, and offer data security. It supports Python, SQL, and markdown cells.

ML Lineage

Snowflake Notebooks

Snowflake offers fully custom ML models and Cortex ML functions. With these, data professionals can easily and securely develop scalable features without having to take data outside snowflake.

Snowpark Container Services is a fully managed solution by Snowflake for deploying, managing, and scaling containerized applications within its ecosystem. It supports running containerized workloads directly within Snowflake, eliminating the need to move data out for processing. Unlike Docker or Kubernetes, it offers an OCI runtime environment optimized for Snowflake, enabling seamless execution of OCI images on Snowflake’s robust data platform.

Snowpark Pandas API

Snowflake Container Services

This platform offers an enterprise-level distributed version of pandas, providing both simplicity and flexibility, coupled with the robust computing capabilities of Snowflake. It delivers the familiar native pandas experience, enabling the execution of open-source machine learning models designed exclusively for pandas. Here's how you can get started

ML Modeling

The Snowpark ML Modeling API integrates popular Python frameworks like scikit-learn and XGBoost to handle data preprocessing, feature engineering, and model training directly within Snowflake.