brianjking

I've been calling this since Friday all over this site and Twitter. It makes absolutely no sense why he's on this board given his direct competition between GPTs and the Revenue sharing versus Poe's creators monetization platform/build your own bot.

The Poe creators monetization is a clear conflict of interest.

Plenty easy, Midjourney.

I guess this is the Open in OpenAI, eh?

Absolutely bonkers.

Absolutely mindboggling that Adam is on the board.

Poe has direct competition with the GPTs and the "revenue sharing" plan that Sam released on Dev day.

The Poe Platform has their "Creators" build your own bot and monetize it, including OpenAI and other models.

less good soundtrack too.

That you can charge people to use, literally almost exactly what GPTs are (minus that those are supposed to be a revenue share from ChatGPT Plus subscribers).

https://poe.com/earnings_tos

Gotta love that Adam D'Angelo who has a clear conflict of interest with Poe to OpenAI's GPTs release from dev day is still on the board and is now somehow the one tasked with leading this negotiation from the boards side.

Well said.

My issue is figuring out how to identify how many concurrent users you can support on average on a given GPU.

Understanding the vram to simply load the weights is easy enough. When you are allowing for something like content generation with varying lengths of input/output tokens, how do you even begin to identify the GPUs you need?

Minus that they are in direct competition to the GPTs and the "app store" and "revenue sharing" that Sam announced on dev day.

Is this book and course helpful when both parties have ADHD? We used to both be medicated. Now only one of us is.

Yeah, I'm having this same problem too. I was paying $20, then switched to the $7 Hacker plan, now have been upgraded again to $20/month Replit Core.

I am torn on what I'll do as I do see value in Replit and really like their product, but not sure I need it at $20/month.

I was paying for the $20 plan for quite some time and switched to the $7/month Hacker plan a month or two ago as it had everything I needed. Disappointed to see my only option is to pay nothing or $20.

Overall, still love Replit, just unsure if I'll continue paying myself. Not because it's bad, but because I don't' necessarily need it.

If this is true, which it very well could be, it's clearly not ok/right. However, this was known for some time. She was tweeting about this in 2021, and this was discussed again in October of this year.

None of that makes sense as to why the board would randomly fire him. I don't think it's this.

It was previously a Google form without any confirmation. As of late October, they moved it to this privacy center that they keep conveniently well hidden.

This page provides confirmation that your request is processed: https://privacy.openai.com/policies

You can opt out of training and keep your history turned on in the privacy center. You fill out a form and indicate your country of residence.

NSFW! flagged!

Great work! Thanks!

One of the weirdest things to come out of the cambrian explosion of AI has been the relationship of these tech giants as they hedge their bets.

Microsoft betting on OpenAI, Databricks, Meta, Oracle, etc.

Google betting on Google/Deepmind, Anthropic, TPUs.

Amazon betting on Bedrock/Amazon/Sagemaker and their ASICs as well as Anthropic.

You, Ethan, & swyx are fantastic.

What is going on here!?

Honestly, I prefer Anthropic as far as just general writing goes. Good luck getting an API key, though.

I'd disagree by far. I use Perplexity as my main search engine and am a paid subscriber. Copilot with GPT-4 or Claude is fantastic.

I love Perplexity.

Ooh, ty! Going to ping you about this.

I'm hopeful that BakLLaVA v2 is equally as good. I also wish / hope that providers using GPT-4 completions to train models would start releasing them with commercially viable licenses if the only issue with the dataset is the use of GPT-4/ShareGPT data.

OpenAI TOS != the law.

https://huggingface.co/SkunkworksAI/BakLLaVA-1

First off, absolutely incredible work, congrats and thank you.

Secondly, do you anticipate Fuyu being made available for commercial access or will it remain NC?

I'd say expertise and compute. With enough compute I can certainly build the largest LLM. Do I have the expertise to build the BEST largest LLM? No.

Kennedy isn't running as a Democrat.

Secondly, without disclosing where you're obtaining the data you're generating embeddings from, this feels pretty dangerous. I hope you're using actual news sources and then providing them in the retrieval.

What was the poor performance that caused you to be let go?

Good luck, friend. I don't think it is too late.