10 AMAZING Llama 3.1 Automations (Fireworks API + Make)

In this video I put the new Meta Llama 3.1 models to the test by having them power 10 AI Automations. I then compare the 405B, 70B and 8B models results side-by-side.

0:00 - Overview of Lama 3.1 models
0:14 - Test scenarios overview
0:47 - API options and costs
1:17 - Llama model pricing comparison
2:01 - Fireworks AI
2:21 - PowerPoint generation
5:32 - Web scraping
9:39 - Large documents
12:42 - Blog article generation
15:36 - Database reports
18:43 - Sentiment analysis
22:15 - Text summarization
24:12 - System message test
24:56 - Prompt injection test
25:31 - Verdict

[pascal]

i didnt understand how you did that, but kudos for doing it and testing and showing us :)

[SouhailEntertainment]

00:00:00 - Introduction and Overview of Llama 3.1 Models
00:00:27 - Test Setup and Deployment Options
00:01:18 - API Pricing Comparison
00:02:12 - Test 1: Generating a PowerPoint Presentation from a Sales Call Transcript
00:05:34 - Test 2: Scraping a Website for Premier League Fixtures
00:09:47 - Test 3: Interrogating a Large Document (Formula 1 Technical Regulations)
00:12:45 - Test 4: Generating a Blog Article on Hydrogen Cars
00:15:32 - Test 5: Running a Custom Database Report
00:19:00 - Test 6: Sentiment Analysis on Tweets Mentioning Southwest Airlines
00:22:37 - Test 7: Summarizing a Web Article on Llama 3.1 Release
00:24:22 - Test 8: Mimicking a Specific Style (Pirate Voice)
00:25:20 - Test 9: Testing for Prompt Injections
00:26:13 - Conclusion and Recommendations

[Toxick]

Great video, some great use cases to help benchmark across the 3 different models!

[HyperUpscale]

I really enjoyed the video - it is very informative and well done 👏

[nickshave]

Really interesting, looking forward to videos on building out Make automations

[WilliamTatum]

use json instead of telling it to just output script - if you constrain it to JSON output you will get MUCH less talk in your code.