Pluralist urban goth, reinforcement_learning(life), transhumanist, d/acc, k/acc, dark maga.
Joined August 2021
- Tweets7K
- Following1K
- Followers213
- Likes43K
7 retweeted
CARNEY just gave $4 million for Gang suppression and disarmament in HAITI 😀
Everyday I’m shocked at how much money is flowing out of the Country. I totally reject it and want my money back. Would love to donate to causes that matter to me
7 retweeted
>fuck CA turned blue I’m moving to CO
>fuck CO turned blue I’m moving to TX
>fuck TX turned blue I’m moving to OK
>fuck OK turned blue I’m moving to FL
>fuck FL turned blue I’m moving to-
>there is nowhere left to go
Running is a temporary solution to a permanent problem.
The 1A, 2A are fundamental to maintaining a free society that actually delivers outcomes to every member of a society, however contradictory it sounds to the false narrative of socialism. Can't wait to show the mathematical simulations on this.
Downtown San Antonio, people signing their hearts out for YMCA, Texas is blue, my ass. Call it cognitive dissonance, or strong priors. But they do feel a little rattled. open.spotify.com/track/4YOJF… – at Olmos Park, TX
9
7 retweeted
Our house of cards economy would topple if consumers could only buy things they could afford
No, it's an incentive of a GPU seller, to have an open model, because it benefits selling consumer GPUs, they're very aware they should be selling both causes because they know no open model can directly compete with proprietary models at scale. This has nothing to do with Jensen, I'd do the same as as a GPU manufacturing CEO.
14
Texas, you feel a little different this time around. More reserved in expressing yourself, a little hesitant, held back, a little unsure, but glad to open of engaged. It feels like an identity crisis? I hope you find yourself through the changes, you've been beautiful, hope these changes can integrate into you a little more seamlessly. – at Alamo Heights, TX
3
This is the kind of product update the world needs for "how's the model coming along?". Truly building open source.
We have dumped JAX/SGLang and rewritten the entire RL and inference stack from scratch in C/C++, which took some extra time.
Training (aka pre-training) is done. Lot of cleanup was needed on the SFT data to get that mostly right. Now, 4.8 is in RL and the evals are starting to climb. We will do a soft rollout to Bot users later this week and see how it goes.
Grok 4.8 will bear no resemblance to 4.7 (thankfully). It is not yet incredible, but, this time, the foundations are built right, so I expect it to evolve to be something special.
7
