Dev tax of the day: typing a polite reply in @cursor_ai ... dang! +$0.08 to the bill (Team account).
Next sprint: rate-limit my manners. eh, no, credit-pooling my manners.
@simonw Honest readme :laugh: BTW what would happen if the previous code is passed on as context to the llm for the dynamic code generation? :thinking: thats nice
@vovahimself The “platonic” idea of a “symbol” for LLMs then?
I love how human realities (like the representation of categories by single representatives, aka platos world of form and ideas) extend to LLMs.
Nvidia driver versions in managed cloud providers for production ML inference are a nightmare.
Not because they are not available, they are even auto-installed. But because of the lack of clear versioning documentation.
I’m the only one struggling with that?
@RudraPrata30055@teknium@maisaAI_ Some code files have the word orchestrator precisely. 🎯 however we find the word orchestrator too generic. What is to orchestrate? Like coordinate different parties for a common goal, and sure, we have to orchestrate the three components named in the tech report
@KaV_2599 Got investment already but Not yet with VCs. AI market is moving fast! We are a small company also. Try to move your project forward to fundraising if have the motivation!
DEMO TIME! Customer service:
Help a customer with a question about an order that did not arrive. This time the customer accidentally did not write the order ID correctly 😯
@aidan_mclau@davipar@maisaAI_ We ran some experiments and the use case of answering a customer email, which involved:
- reading the email
- finding the order in a big (50MB size) CSV
- understanding the order details
- composing the email
- actually sending the email
Had a cost of ~0.2$ in under a minute.
@Footballen93999 I mean that it is going to be difficult to identify the AGI. Even if there is a consensus on the “what is it” the identification problem exists and will exist. How can we identify te first AGI? 🤷♀️
Worth your question for clarification btw 💯
@DrewHawkswood@WilliamBakst Yeah, yeah, there are benchmarks that will last years. There are benchmarks that don’t. And there is stuff that is not benchmarked. This is the point, we need more benchmarks that last years. 🎯