How to Accelerate AI Training With Programmatic Data Labeling: Snorkel Flow Demo
Predictive models, aligning LLMs, and fine-tune RAG applications all require labeled data. However, enterprise AI initiatives quickly become blocked due to the excessive time and cost needed to label data manually.
The solution is programmatic data labeling, which provides enterprises with an efficient way to capture knowledge from SMEs and then use it to automate data labeling at scale. It accelerates AI development time from months to days.
In this webinar, Snorkel AI Senior Applied Machine Learning Engineer Nazanin Makkinejad, explains how enterprises can not only accelerate data labeling, but quickly iterate, adapt, and improve label accuracy via AI data development.
In addition, she shows the following real-world use cases:
Classifying chatbot utterances for a bank
Extracting procedures from docs for a pharma company
Extracting drugs and adverse events from medical notes
See more Snorkel Flow demos here: • Snorkel Flow Demos: See How it Works!
Timestamps:
00:00 Introduction
01:36 Motivation for Programmatic Labeling
05:35 Challenges with Manual Labeling
08:34 Introduction to Programmatic Labeling
08:44 Overview of Snorkel's Origins
08:51 Applications of Programmatic Labeling
08:59 Explanation of How Programmatic Labeling Works
12:03 Labeling Functions and Their Creation
14:46 Steps in Programmatic Labeling
17:52 Real-World Impact of Programmatic Labeling
19:18 Demonstration Introduction
20:27 Demo: Classifying Chatbot Utterances
24:12 Demo: Training a Model
28:34 Demo: Creating Labeling Functions
31:29 Demo: Using Embeddings for Labeling Functions
39:31 Demo: Drug and Adverse Event Extraction
44:40 Summary and Conclusion
45:07 Q&A Session
#enterpriseai #programmaticlabeling #datacentricai
