THE SNOWFLAKE CORTEX UMBRELLA

THE SNOWFLAKE CORTEX UMBRELLA

Curious about Snowflake Cortex and what it can do for you? Think of Cortex as your research assistant that helps you sift through information, find relevant details, and answer your questions to make sense of your structured and unstructured data. Your own personal data interpreter at your fingertips. Let’s break it down and find out where to get started!

CORTEX ANALYST: INTERACT WITH STRUCTURED DATA, CONVERSATIONALLY?

Cortex Analyst is built using Meta’s Llama and Mistral models to streamline the development of self-service analytics with more flexibility than standard dashboards to answer your data questions.?

Say you’re an analyst trying to look at marketing campaign data. You want to explore and understand metrics across various marketing campaigns across channels. Cortex Analyst can help you receive direct answers about your data without writing SQL. Build a chat application using Streamlit and use natural language to receive answers like “what promotional channels drive the most views to my content?” You can create a semantic model to make sure your data is aligned with your target task (in this case, you would want to make sure you limit the tables and columns to those most applicable).?

Try it today! Cortex Analyst is free until November 15, 2024.?

CORTEX SEARCH: ASK ENTERPRISE DOCUMENTS YOUR QUESTIONS

Index and query your unstructured data and documents using Cortex Search for data ingestion, embedding, retrieval, reranking, and generation. Find the answers you’re looking for across a large number of documents.?

When reviewing customer service tickets to see if there are common sources of feedback related to frequent issues, run a SQL command to create a database and dataset by inserting values such as ‘billing issues,’ ‘payment,’ and ‘overcharges’ to retrieve all the relevant data using high quality, “fuzzy” searching.?

CORTEX FINE-TUNING: SECURELY CUSTOMIZE LLMS WITH SERVERLESS FINE-TUNING

Improve task accuracy by making adjustments to models. Fine-tune by calling an API or SQL function. Select the base model (can be from Meta or Mistral AI), provide your training data set, and Cortex will take care of the rest.?

Try it yourself: https://developers.snowflake.com/solution/llm-serverless-fine-tuning-with-snowflake-cortex-ai/?

See example below for the “support_tickets” dataset.?


SNOWFLAKE CORTEX GUARD: FILTER AND FLAG TO ENSURE MODEL SAFETY

Snowflake takes AI safety seriously. Within the Snowflake Cortex AI COMPLETE function, add ‘guardrails’: true’ to activate Cortex Guard.

Cortex Guard is currently powered by Llama Guard 2 from Meta, evaluating a response before the output is returned. Llama Guard 2 protects against violent crimes, non-violent crimes, privacy, intellectual property, hate, and more.?

SNOWFLAKE COPILOT: YOUR LLM-POWERED ASSISTANT

Create a new worksheet or open an existing one and select “Ask Copilot” in the bottom right corner. Select a database and schema, type your question, and watch the magic happen. Snowflake Copilot can provide you SQL statements that you can then run directly in the worksheet.?

A hospital system is wanting to identify patient flow and resource usage to highlight inefficiencies or bottlenecks. Once a table and schema are selected, you can ask Snowflake Copilot questions such as, “What are the peak times for emergency room visits?” or “How do wait times vary by department or time of day?”

You can even set custom instructions up to a 2,000 character limit, in plain text English. Start your conversation with Snowflake Copilot today!?

DOCUMENT AI: LLM-POWERED DATA EXTRACTION

Process documents of various formats and extract information from text-heavy paragraphs and graphical-form content. Turn unstructured data from docs into structured data in tables.?

  • Zero Shot Extraction: the foundation model can locate and extract information specific to a doc type, even if it’s never seen the document before
  • Fine Tuning: tune the Snowflake Arctic-TILT model to improve your results by training the model on your documents for your use-cases

ELEVATE YOUR SKILLS THROUGH LEARNING

BUILD: THE DEV CONFERENCE FOR AI & APPS

Register for free today for BUILD: The Dev Conference for AI & Apps. Join developers, data scientists, engineers, and all data professionals for three days of product announcements, technical “how-to” sessions, and hands-on labs in a region near you.

INTRO TO MODERN DATA ENGINEERING ON COURSERA

Enroll in the newly launched data engineering course on Coursera. Learn how to build modern and continuous data pipelines with Snowflake and level up your understanding of ingestion, transformation, and orchestration.?

UNIFY YOUR GOVERNANCE STRATEGY WITH SNOWFLAKE HORIZON

Join this instructor-led lab on Sep 25 to learn how Horizon dashboard can securely monitor and provide visibility into your sensitive data within Snowflake.


Qi Sun, Ph.D.

I want to build Tao following Nature society. Please hire me.

1 个月

The artificial magnetic waves cause the increase of radiation of the Sun. It is the same as the electricity causes blackbody radiation. AI is just an advanced tool. AI is a big liar which causes more and more people losing their jobs. AI is creating more and more artificial magnetic wave applications and wasting more and more energy. So, we need to reduce the artificial magnetic wave applications, especially wireless applications. We are in deep fake and evil world right now. Modern science is deep fake, modern technology is evil. The more fake knowledge there is, the more reactionary it becomes. 5:41 @ https://www.youtube.com/watch?v=Rw3sfhW3Bt0

回复
Shin Kae Huang

Area Leader in Taiwan & Hong Kong | Entrepreneur Spirit| Digital Transformation | Cloud, Data and AI/ML Analytics | Self-Driving & Learning, Capability, Collaboration and Communication

1 个月

That’s great to make data and real AI together! Not just a slogan or roadmap! ??

回复

Hot topic, thank you for highlighting! ????

回复
Bob Barone

Transformational Operations Executive | Innovative Leader in Healthcare & Technology | Scaling High-Growth Organizations through Continuous Improvement | Client Implementations & Engagement

2 个月

Amazing how far things have advanced. So interesting!

回复
Naveen Kumar M

AVP Software Engineer | C# Corner MVP ??

2 个月

Good to know

回复

要查看或添加评论,请登录

社区洞察

其他会员也浏览了