Interesting results! For me, Claude 3 is so much better than GPT-4, to the point where I stopped using GPT
GENERATIVE AI
-
AI-Powered Newsletter Workflow: From 7 Hours to 2
By
–
Just created a 28-page AI SOP for my newsletter It’s glorious. Original workflow: write everything from scratch, probably 7h of work Now: dictate newsletter into Otter AI, create draft with new SOP doc (notion) and AI (combo of GPTs and Claude), edit and add links, max 2h
-
AI Interpretability Challenge: Emergent Properties Beyond Builder Understanding
By
–
It’s D. Even the people building AI don’t fully understand what it’s doing. That’s why so many AIs have “emergent properties.” I also think C is somewhat true. Those building it aren’t that interested in explaining it. Unless you’re @karpathy
. -
New Extraction Guides for Structured Data from Unstructured Text
By
–
New extraction guides Reliably extracting structured data from unstructured text is one of the killer use-cases for LLMs. It's a fantastic way to bridge the gap between LLMs and traditional APIs and systems. We've put a lot of work on this lately, including an open
-
Beyond Models: Crafting Compound AI for Concrete Value
By
–
Check out my latest article: Beyond the Models: Crafting Compound AI for Concrete Value https://
linkedin.com/pulse/beyond-m
odels-crafting-compound-ai-concrete-value-ganesh-padmanabhan-isnpc
… via @LinkedIn -
Upload Designs and Choose Style Models for Custom T-Shirt Mockups
By
–
Yes, simply upload your designs and select a style model. You can even upload designs and specify a person, and then you will have that person wearing the t-shirts. For that, use the training type person!
-
Fine-tuning LLMs: Democratizing AI Performance with Smaugh
By
–
Fine-tuning can change and customize your LLMs to your particular domain, allowing even the GPU poor to improve base models by 10-15%. Additionally, it’s the way for open-source AI to match GPT-4 performance. In the past month, we at Abacus AI have been announcing Smaugh
-
Testing Megaprompts on Claude-3 and Gemini Ultra Models
By
–
Whenever a new foundational model comes I like to run my best megaprompts in it. Gemini Ultra has been awesome at this. So is Claude-3. But Claude-3 has a huge context window. Praying Gemini Ultra will get one soon. I need those tokens!
-
Breaking Confidentiality Paradox: Data Protection in GenAI
By
–
Join Shaun Hillin, Head of Solutions Architecture at Cohere as he speaks on the 'Breaking the Confidentiality Paradox' panel at @sxsw on March 11 10:00 AM CET. Gain exclusive insights on how data protection is integral to enabling a secure GenAI roadmap.
-
Yi-34B-200K Achieves 99.8% Performance in Long Context Test
By
–
Yi-34B-200Khas enhanced the long text capability. In the Needle-in-a-Haystack test, the performance rises from 89.3% to 99.8%. @01AI_Yi continues to pretrain the model on 5B tokens long-context data mixture and demonstrates a near-all-green performance!
https://
github.com/01-ai/Yi?tab=r
eadme-ov-file#news
…