Hacker News

Subscribe to Hacker News feed
Hacker News RSS
Updated: 29 min 17 sec ago

AI Rendering Playground for Architects

Thu, 10/31/2024 - 9:08am

Article URL: https://neolocus.ai

Comments URL: https://news.ycombinator.com/item?id=42006534

Points: 1

# Comments: 1

Categories: Hacker News

Ask HN: Next Token to Next LLM

Thu, 10/31/2024 - 9:04am

I was thinking that why are we investing so much effort on predicting next token when we can basically change our approach to predicting next version of the LLM itself i.e. a LLM which generates new LLM with weights - until it maxes out all benchmarks

Comments URL: https://news.ycombinator.com/item?id=42006512

Points: 1

# Comments: 0

Categories: Hacker News

How Organizations are handling agent microservice model?

Thu, 10/31/2024 - 9:00am

We are central framework team which provides agent to multiple product teams to take of common functionality.

We use Agent-Server, our provided agent will call our central servers to fetch data. We are actively working on codebase, we release server changes every week twice and agent changes monthly once, and as when needed if bug arises.

I see this maintaining and working in such central team, we have to be more cautious and safe while making changes, still some edge cases missing and triggering bugs. We do have vast amount of test cases and integration tests. still as there are 100+ apps, maintaining integrity seems so difficult.

We do have staging area and prod, it safe guards tons of errors. but still I think we could improve some processes.

I would like to know details about how big orgs manage their agent server models.

Comments URL: https://news.ycombinator.com/item?id=42006478

Points: 1

# Comments: 0

Categories: Hacker News

Luna- SSR React Wiht Go

Thu, 10/31/2024 - 8:59am

Article URL: https://github.com/Djancyp/luna

Comments URL: https://news.ycombinator.com/item?id=42006470

Points: 3

# Comments: 1

Categories: Hacker News

Ask HN: How much are you spending on AI?

Thu, 10/31/2024 - 8:16am

Hello there!

I'm working in a company where we use an AI provider (Anthrophic) and I'm curious how much are some of you folks spending on it, be it OpenAI, Anthrophic, etc., (when the company objective per-se is not AI) on features to support the core functions of the product.

Just for reference, yesterday I spent almost $10 while developing in my local machine, and it felt like "a lot".

PS: I understand that this is a question that the business needs to address too, i.e. understanding the margins of the product taking into account all costs. Which might imply that sometimes $1 per day is too much, or $1k per day is cheap.

Consider this Ask HN as curiosity driven :D

Comments URL: https://news.ycombinator.com/item?id=42006077

Points: 1

# Comments: 0

Categories: Hacker News

All Talks from the Matrix Conference

Thu, 10/31/2024 - 8:09am
Categories: Hacker News

Pages