As always, when you try to analyze the capabilities of these models, please look at the datasets they’ve been trained on! https://
x.com/hahahahohohe/s
/hahahahohohe/status/1765435151817830834
…
@clementdelangue
-
Analyzing AI Model Capabilities Through Training Data Examination
By
–
-
Anthropic Shares Claude System Prompt for Transparency
By
–
Great transparency from @anthropic @AmandaAskell sharing the system prompt of the current API of Claude. It helps us understand why it works the way it does. My assumption why the generated text sounds much more human than previous systems:
#1 it's more heavily trained on human -
AI Training Makes Systems Sound More Human, Not Actually Human
By
–
In short, AI doesn’t « get more human », the developers of AI make it sound more human thanks to its training (this is a technology system). https://t.co/NwnX6aaNZH
— clem 🤗 (@ClementDelangue) 5 mars 2024In short, AI doesn’t « get more human », the developers of AI make it sound more human thanks to its training (this is a technology system).
-
Claude is Code and Algorithms, Not Human Intelligence
By
–
AI/Claude is cool of course but it is not human (despite its name), it's a bunch of lines of code and probabilistic algorithms trained on datasets behind an API to take an input and generate an output. Let's please all remember this for healthier debates and less fear-mongering!
-
Claude is Code: Debunking AI Misconceptions for Healthier Discussion
By
–
AI/Claude is cool of course but it is not human (despite its name), it's a bunch of lines of code and probabilistic algorithms trained on datasets behind an API to take an input and generate an output. Let's please all remember this for healthier debates and less fear-mongering!
-
Anthropomorphization in AI model learning descriptions raises concerns
By
–
Very very cool work and don't want to take away anything from the magnitude of your achievements but the level of anthropomorphization of your tweet makes me uncomfortable. Like for example, I feel like it's misleading to say "help these models learn more and more" and refer to
-
Tech Outages Inevitable Cost of Building Successful Products
By
–
You obviously want to minimize them but there's no way to build a successful tech product without outages at some point. Anyone who built successful products understands that. Sending support to the instagram & meta engineering teams working hard to get it back up under
-
AI Evaluation Bug Discovered During Testing
By
–
And I thought only AI evaluation could be wrong Given the feedback, it looks like there’s been a bug (I’m not a pro-cyclist) so I’ll try to retake the test soon!
-
AI Evaluation Bug Discovered During Testing
By
–
And I thought only AI evaluation could be wrong I didn’t fill my weight but given the feedback, it looks like there’s been a bug so I’ll try to retake the test soon!
-
Text-to-Video AI Joining Generative AI Revolution
By
–
So excited to see text to video joining the party. You can learn more about this specific task here including with some open-source models and datasets: https://
huggingface.co/tasks/text-to-
video
…. AI is going to transform every single modality, domain and tasks of technology building!