Some interesting examples: In this image and a few others, CLIP appears to associate bare skin with "embarrassment". LLAVA and Captions + GPT don't, seeming to reason over the location and context.
GENERATIVE AI
-
Character.AI Announces c.ai+ Subscription Giveaway
By
–
http://
Character.AI | c.ai+ GIVEAWAY Thank you so much for your support! As we reach each member milestone on Discord, we will be selecting lucky winners to receive a c.ai+ subscription! Join the #giveaway at http://
discord.gg/characterai -
Unpaid AI Work: Volunteers vs. Financial Incentives
By
–
I very much doubt anyone is paying them for this. Certainly all my work in the field is as an unpaid volunteer. There's piles of money available on the other side, so anyone in this for $$$ should focus there.
-
Openness and Transparency as AI Risk Mitigation Strategies
By
–
Other things could *also* mitigate those harms. But openness, transparency, and broad access do seem likely to be helpful towards mitigation. You could quite reasonably argue that some firms today are "openness-washing" – but that doesn't make the statement wrong either AFAICT.
-
AI Companies Revolutionizing Content Creation and Design Innovation
By
–
Looking for #AI companies that excel in content creation? These innovative companies are revolutionizing the creative landscape with their cutting-edge AI technologies. From writing to design, they're transforming the way we create captivating content.#Innovation Via @ingliguori
-
AI as Creativity Engine for Widespread Productivity Improvements
By
–
"This is a time when we need productivity improvements like never before. AI is an engine for creativity that is going to be widely available to everybody" Thanks to Fiona Bruce and the @bbcquestiontime team for inviting me on. https://
bbc.co.uk/iplayer/episod
e/l0056fxm/question-time-question-time
… -
Adaptive Prompting Approach Improves LLM Few-Shot Performance
By
–
Introducing a new approach for adaptive prompting of #LLMs that train with unlabeled samples + pseudo-demonstrations generated by the model itself to close the gap between few-shot and 0-shot performance on reasoning, NLU and language generation tasks. → https://t.co/ZtSEZOxNCc pic.twitter.com/5H7YE0usc3
— Google AI (@GoogleAI) 2 novembre 2023Introducing a new approach for adaptive prompting of #LLMs that train with unlabeled samples + pseudo-demonstrations generated by the model itself to close the gap between few-shot and 0-shot performance on reasoning, NLU and language generation tasks. → https://
goo.gle/3tY00IB -
LAION Dataset Ethics in Stable Diffusion Model Training
By
–
What she's doing is relatively standard then. Don't see how it's much more ethical… even the first released SD models were LAION.
-
Understanding LLM Word Prediction: Key Knowledge for AI Progress
By
–
What makes this knowledge pivotal is its relevance to how LLMs predict the next word of a sentence or respond to inquiries. Equipped with this understanding, you'll be able to make strides in the AI space with ease.
-
Understanding LLM Architecture: Deep Dive into Transformers
By
–
If you want to have a comprehensive look at how LLMs function and understand their architecture better, don't wait, dig into the full article! https://
louisbouchard.substack.com/p/how-do-llms-
work-the-transformer
…