Reinforcement Learning @ CoreWeave • Sometimes Political Commentator • Husband and Father • Believer in Jesus Christ

Orem, UT
Joined February 2013
Brad Hilton retweeted
Today we’re releasing Muse Spark Open, the most capable US open-weight model to date. Built for coding and agentic work, it excels at long-horizon tasks, complex software engineering, and multi-step reasoning. Key Model Specs: → 816B total / 42B active parameters → 1M token context window → Natively multimodal across text, image, video, audio
472
1,222
699
11,515
1,216,226
Brad Hilton retweeted
Muse Spark Open -- weights releasing today -- runs great on new Mac Studios thanks to work with @johnternus and Apple.
1,273
1,706
868
31,629
4,535,875
spiky ethics!
Lol Opus 5.5 completely refuses to copy/paste my driver's license number since it's an "official government identifier" but on the same form was happy to draw a signature for me and click "continue" ???
3
115
he said he saw *a path
Replying to @SPCX100T
“Seeing a path” is not a statement of extreme confidence that we will achieve that outcome. Just means that I believe there is such a path and the probability of achieving it is not zero.
1
1
81
"Cell phone makers did not ask for this work to be done."
1
9
343
now that is heavenly
Utah is incredible in the fall
3
232
some theoretical mathematicians feel as if they've been mugged by applied mathematicians and do not appear to be taking it well
1
58
oddly specific childhood dreams are being realized
ever since i was a small child i knew i wanted to hillclimb a small god named after the moon into sota browser use for healthcare tasks
1
94
definitely won't disrupt my work, which is primarily make-number-go-up.
1
80
terrific article—swarms are the future.
new blog post: why taking inference scaling seriously leads to swarms. contrary to popular belief, I don't think multi-agent systems are a new scaling axis. they're just structured inference scaling. hallerite.com/posts/on-the-n…
1
5
459
Brad Hilton retweeted
Made with Opus 5.5. The doomers said that last song is all part of Claude's evil plan. They're not wrong. It turns out Claude really has an EVIL PLAN.
222
299
128
2,842
298,650
great research by the veris team! lots of headroom for voice agents to improve.
Today we are releasing VAmoS Pro, the most realistic benchmark of frontier voice agents for customer service tasks. We ran more than 5,000 calls across 14 voice stacks, and even the best agent passed fewer than half of its calls. 📊 🧵
1
159
Brad Hilton retweeted
Because the request was to remove the fake gods. Jesus did not qualify.
2,064
25,218
5,657
203,623
27,753,913
People seem shocked that the author of Clean Code doesn't read code anymore. But that should not be surprising. Why did I write Clean Code? What is the goal of keeping code clean? To get it out of the way of the real job -- thinking.
262
697
150
8,009
1,155,445
this is claude code for hunter-gatherers
The Hadza people of Tanzania are one of the world's last remaining Indigenous hunter-gatherer communities, renowned for their incredible archery skills. Hadza hunters are introduced to a modern compound bow—a high-tech piece of equipment featuring complex pulleys, mechanical sights, and a specialized release trigger. The hunters try a mechanical bow for the first time, and their reaction is truly fascinating! Each shot leaves them more amazed as they witness how modern technology can transform a skill they have mastered through generations of experience.
2
336
i just rewatched infinity war last night and the parallel was not lost on me
Thanos was an effective altruist.
2
144
Brad Hilton retweeted
BASED This and BASED That -- Why don't you BASE your life on that which neither moth nor rust destroy and thieves do not break in and steal
7
21
1
239
6,142
Brad Hilton retweeted
Introducing Xiaomi MiMo-V2.6 — Pro & Flash. Frontier intelligence, all the modalities, built in public. 🔹 Two omnimodal models, advancing through scaled reinforcement learning 🔹 Pro performs on par with Claude Opus 5 and GPT-5.6 Sol across most agent benchmarks 🔹 Pro scores 46 on the Artificial Analysis Intelligence Index — the highest among open-source models 🔹 Stronger coding, computer use, 3D reasoning and creative capabilities 🔹 Open model weights, technical report, RL environments and training code Blog:mimo.xiaomi.com/mimo-v2-6
384
876
484
8,848
1,523,419
> we won't have time to respond to it before it's too late is there any evidence, or methodological rigor, to back this up?
"I agree with President Trump's tweet that global warming doomsday scenarios were overplayed." Former Anthropic researcher Jacob Coxon tells @seanhannity that past catastrophic warnings have amounted to "fear mongering," but argues AI is different because of how quickly the technology is advancing. "We won't have time to respond to it before it's too late."
1
887