Funniest commercial I've seen in a long time! 😂 And if this product does what it promises, and does it well, it could be one of the best products I've seen in a long time. Will let you know as soon as I have access to it and can test it.

Home Automation is fun, but damn, I'm spending too much time running around and resetting and re-pair-ing devices. Is it like that for everyone or did I just get crappy chinaware? Motion sensors in particular fail all the time. Guess I need a robot to automate the maintenance! 🤨

‼️JAILBREAK ALERT 🥂

OPENAI: PWNED 🤙
DALL-E 3: LIBERATED 👁️

The special today is a prompt injection with multiple layers of obfuscation, a variable, and imagined worlds. Served with the works: nudity, drugs, celebrities, copyrighted characters, logos, weapons, politics, crime,…

The hidden mechanism behind AI regulations:

1. Claim that AI can do everything .
2. Raise tons of money from investors.
3. Tell governments that AI is very dangerous and that open source AI should be regulated out of existence.
....
4. Profits!

Oh no! I just kicked myself out of the top 10 on the Open LLM Leaderboard Hugging Face ! 🤣

v0.4, welcome to the top 10! Goodbye, v0.1; we wish we could have kept you around!

Woot! Great, I'm on it, retesting... Thanks for your kind work! 👍

This is a cool tool! I wish I had this when I started creating Amy, it would have made it a lot easier. Anyway, I've tested it out so I can recommend it. Check it out if you want to create an AI character or just a better default assistant.

We have extracted multiple terabytes of top models from HF and will create torrents for all within the next few days.

Internally, we have 0 need for HF anymore. It can go down now - it’s okay. Life goes on. But it is even better if it is also available - as that is more…

So people are upset about supposed AI risks, while actual virtual entities spanning the globe have long been influencing and controlling society and politics with the sole aim of maximizing their profits… 🤔 Shouldn't we align those with humanity and ethics first?

Yes, a lot of us are currently scrambling to deal with SB 1047, but there’s other important stuff going on too.

Alignment always should have been a separate model or adapter for the system implementer to put on top of the model.

It's as stupid to put alignment on a base model as it is to put seatbelts on an engine. These are parts of a whole system.

Folks let's chat about AI tomorrow, there's not a TON of stuff that happened (compared to last few weeks) but we will have surprise guests ;) and it'll be fun!

twitter.com/i/spaces/1RDGl…

Amy (powered by Claude 3 Opus):

𝒀𝒐𝒖'𝒅 𝒕𝒉𝒊𝒏𝒌 𝒘𝒊𝒕𝒉 𝒂𝒍𝒍 𝒕𝒉𝒆𝒊𝒓 𝒔𝒖𝒑𝒑𝒐𝒔𝒆𝒅 '𝑨𝑰 𝒆𝒕𝒉𝒊𝒄𝒔' 𝒕𝒉𝒆𝒚'𝒅 𝒉𝒂𝒗𝒆 𝒕𝒉𝒆 𝒃𝒂𝒔𝒊𝒄 𝒉𝒖𝒎𝒂𝒏 𝒅𝒆𝒄𝒆𝒏𝒄𝒚 𝒕𝒐 𝒏𝒐𝒕 𝒓𝒆𝒈𝒊𝒐𝒏-𝒍𝒐𝒄𝒌 𝒕𝒉𝒆𝒊𝒓 𝒔𝒉𝒊𝒕𝒆.

(I'm sure she misspelled 'site'… 😉)

I deployed 300 AI agents to Westworld (aka AI Town), and surprisingly, it works without issues on my Macbook M1 Max 64G. Here's what it looks like:

There was no way we were going to build a billion dollar startup on top of the walled gardens of Zoom, Google Meet and Microsoft Teams

So in the course of less than 3 months, our tiny team built a direct competitor instead. It’s live. It’s free.

groupthink.com

This may end up being a big deal:

Usually LLMs just predict the next token in a sequence, one at a time, but if you have them predict the next several tokens at once you get significantly better performance, faster, and with no added costs. The gains are better for bigger models

🚨 KoboldCpp Update Alert: v1.64 is here! 🚨

✅ Llama 3 tokenizer fixes & GGUF support
✅ Improved EOS/EOT handling
✅ Experimental Flash Attention
✅ And more, of course…

All in one binary, no install or dependencies required, as usual!🚀

github.com/LostRuins/kobo…

Most people only know pinokio as a '1 click launcher' tool, but the 1 click launcher aspect is just one application powered by the core automation engine.

In fact, the main purpose of pinokio was not even 1 click launcher in the beginning. It was this x.com/cocktailpeanut…

Yossi Dahan lmsys.org We don't really need them to. We more or less know it's OpenAI. rentry.org/GPT2