Explorar

New day, new checklist. - The launch of this super work AI that I have been working on so long has been delayed for four months. It's eaten every ounce of hope in me. But I'm going to launch it anyway, not going to delay it anymore. We're going to talk about it, tell the world what we are building and where we're heading towards, and just put it on a waitlist if have to…. This way, I can just get the launch and the hype out of the way and focus on trying to get this product to the finish line Let’s see how it goes

CEO @Plutio_app - on a mission to kill busywork with the world’s first Super Work AI. Acquired in a $10M talent deal. Ex–Head of Brand @ClickUp.

Leo Bassam 🤌🏼

Sun Dec 07 00:48:15

There's a surprising ideological dimension to Rnj-1, an opposition to DeepSeek and OpenAI swaying the field towards RL. They argue, in effect: «you haven't learned how to walk yet; betting the house on running RL is foolish». PI can do RL. I wonder where Essential will go.

In my book, Essential joins the ranks of DeepSeek and Prime Intellect as an «ideological lab», proposing its own principled research program for advancement of open intelligence. Maybe we can add Arcee as #4, with their merging focus. I celebrate having more schools.

Teortaxes▶️ (DeepSeek 推特🐋铁粉 2023 – ∞)

Sun Dec 07 00:46:19

Might validate my intuition that a multi-layered reasoning style can provide great anchors for RL climbing.

Artisanal baker of reasoning models @pleiasfr

Alexander Doria

Sun Dec 07 00:46:13

And new major open source LLM project unlocked. Not only addressing critical scalability issues for MoE research, but also better data pipelines.

(I was just setting up torchtitan this week for a mixture of baguettes)

Alexander Doria

Sun Dec 07 00:40:38

"Have you tried switching it off and on again Mechanisms and therapeutic prospects of resetting homeostatic set points in medicine and neuropsychiatry" a levin preprint strikes again: https://t.co/GpJWvWKgy4

third thing founder https://t.co/jZh799yNH4 / personal https://t.co/IdaJwZJCXm

near

Sun Dec 07 00:39:41

Remarkable. India and Bangladesh are very similarly concerned; clearly both are more likely located on OpenAI's servers. And endless rehashing of Chinese fishing malfeasance, videos 4-8 years old surreptitiously mixed with new. Who's running this grift?

We're in a race. It's not USA vs China but humans and AGIs vs ape power centralization. @deepseek_ai stan #1, 2023–Deep Time «C’est la guerre.» ®1

Teortaxes▶️ (DeepSeek 推特🐋铁粉 2023 – ∞)

Sun Dec 07 00:39:06

Newest first — browse tweet threads

Explorar

Newest first — browse tweet threads

There's a surprising ideological dimension to Rnj-1, an opposition to DeepSeek and OpenAI swaying the field towards RL. They argue, in effect: «you haven't learned how to walk yet; betting the house on running RL is foolish». PI can do RL. I wonder where Essential will go.

Might validate my intuition that a multi-layered reasoning style can provide great anchors for RL climbing.

And new major open source LLM project unlocked. Not only addressing critical scalability issues for MoE research, but also better data pipelines.

"Have you tried switching it off and on again Mechanisms and therapeutic prospects of resetting homeostatic set points in medicine and neuropsychiatry" a levin preprint strikes again: https://t.co/GpJWvWKgy4

Remarkable. India and Bangladesh are very similarly concerned; clearly both are more likely located on OpenAI's servers. And endless rehashing of Chinese fishing malfeasance, videos 4-8 years old surreptitiously mixed with new. Who's running this grift?

Explorar

Newest first — browse tweet threads

Explorar

Newest first — browse tweet threads

There's a surprising ideological dimension to Rnj-1, an opposition to DeepSeek and OpenAI swaying the field towards RL. They argue, in effect: «you haven't learned how to walk yet; betting the house on running RL is foolish». PI can do RL. I wonder where Essential will go.

Might validate my intuition that a multi-layered reasoning style can provide great anchors for RL climbing.

And new major open source LLM project unlocked. Not only addressing critical scalability issues for MoE research, but also better data pipelines.

"Have you tried switching it off and on again Mechanisms and therapeutic prospects of resetting homeostatic set points in medicine and neuropsychiatry" a levin preprint strikes again: https://t.co/GpJWvWKgy4

Remarkable. India *and* Bangladesh are very similarly concerned; clearly both are more likely located on OpenAI's servers. And endless rehashing of Chinese fishing malfeasance, videos 4-8 years old surreptitiously mixed with new. Who's running this grift?

Remarkable. India and Bangladesh are very similarly concerned; clearly both are more likely located on OpenAI's servers. And endless rehashing of Chinese fishing malfeasance, videos 4-8 years old surreptitiously mixed with new. Who's running this grift?