Advanced Text Generation and Analysis with LLMs

[BaDshaH]

Published 2/2024
MP4 | Video: h264, 1920x1080 | Audio: AAC, 48 KHz
Language: English | Size: 103.84 MB | Duration: 38m 17s

You've got a basic understanding of LLMs and may have built a proof of concept or two. This course will demonstrate deeper engineering techniques and help you build a strategy to assess and improve the fluency, relevancy, and precision of your LLMs.

Moving beyond a hobby project with a large language model (LLM) to a real-world application often requires advanced techniques to ensure your solution is production-ready, and that it will remain contextually effective under scrutiny and scale.

In this course, Advanced Text Generation and Analysis with LLMs, you'll gain the ability to assess, monitor, and guide the text generation of your LLM.

First, you'll look at advanced techniques to ensure your LLM maintains relevancy and coherence within a domain, increasing its fluency while maintaining context.

Then, you'll explore additional techniques to increase your LLM's ability to identify and generate intent, sentiment, and tone in both pure natural language processing environments as well as multimodal applications such as video, image, and audio AI applications.

Next, you'll see how to guide the output of your LLMs, researching prompt engineering - an especially critical concept with conversational models or question/answer applications - as well as demonstrating techniques to guide both a model's diversity and style.

Finally, you'll look at the various methods researchers have created to assess and monitor the effectiveness of an LLM model, introducing concepts such as precision and recall as well as popular metrics such as BLEU and ROUGE.

By the end of this course, you'll be prepared to utilize large language models in a real-world setting, with increased confidence in your model's ability to accurately perform within the context of its intended application.

Homepage

[To see links please register or login]

[To see links please register or login]