People that are bullish on AI focus on the value of frontier tasks, while those bearish focus on lack of diffusion for basic tasks. But if anything that should make them MORE bullish!
0
0
6
578
Someone today referred to RL tasks as "non-fungible commodities", most apt definition of the space I've heard yet
4
0
21
1,837
🚀🚀🚀
Proud to be a partner in the @FireworksAI_HQ Frontier Ecosystem, announced today at Forge
Preference Model + Fireworks: train your own models on RL environments that are hard, verified, and built to resist reward hacking.
Come build with us: preferencemodel.com/blog/preferenc…
0
0
2
550
Mental health is one of the least discussed benchmarks today, and so it is fantastic to see the @tryterac team partnering with @Prosaic_int here! As personal agents continue to grow, the world needs MUCH more work here
It was great to work with @Prosaic_int on their new benchmark.
Whether a model notices and responds safely to suicide risk spread across a user’s history is one of the highest-stakes questions in alignment.
Thank you to the @tryterac expert clinicians who helped outline what clinically correct responses look like!
0
2
11
945
Adapt or die. Whether or not you sell your internal data or license your content, change is coming. Companies can either profit off this, or put their heads in the sand
It interesting thought exercise is whether the business model for these companies just becomes being a data vendor of a very particular type.
0
0
3
745
Ryan Wexler retweeted
Genuinely the best conversations I’ve had all year on where RL and human data are going.
Thank you to our speakers, @SignalFire, @contralabs and everyone who joined us. See you at Convergence 2!
Thank you to our speakers, @SignalFire, @contralabs and everyone who joined us. See you at Convergence 2!
Yesterday we hosted Convergence: RL Summit bringing together researchers, industry leaders, and operators to talk about the frontier of model development and where things are going.
1
2
21
1,281
Ryan Wexler retweeted
Fantastic day at @tryterac’s inaugural RL Summit - great mix of eng, research, and applied folks all in one room.
Thanks to @zacbakerr and @RyanMWexler for putting together such a sharp group, and for letting me share what we’re seeing at @chakra_ai. Excited to return next year!
Thanks to @zacbakerr and @RyanMWexler for putting together such a sharp group, and for letting me share what we’re seeing at @chakra_ai. Excited to return next year!
4
7
31
1,280
Ryan Wexler retweeted
Yesterday we hosted Convergence: RL Summit bringing together researchers, industry leaders, and operators to talk about the frontier of model development and where things are going.
9
10
50
5,328
Two dinners, one running event, and a whole conference later we have wrapped up the first annual Convergence: RL Summit co-hosted by the @SignalFire and @tryterac teams! terac.com/events/converg…
1
1
37
1,603
But most importantly I want to thank the audience for taking times out of their day to join us. We had a fantastic group ranging across researchers at the frontier labs, current/future founders of RL companies, SPLs, and those newer to the space.
1
0
3
83
We'll be releasing recordings of the event over the next few weeks, and until then please reach out to myself, @zacbakerr, or @JackBlair87 if interested in chatting all things RL and data. It is so exciting seeing all the interest here, and looking forward to running it back!
2
0
4
112
Doors are open for the Convergence: RL Summit! Come by for breakfast, and talks start in 30 minutes!
0
0
22
1,251
Fun fact: Per ChatGPT Michael Dell owns 24x more shares of Broadcom than their own CEO Hock Tan after selling off VMware in 2022!
SCOOP: Broadcom CEO Hock Tan was getting impatient.
He needed lenders to commit $40 billon in the largest AI debt deal ever—so Anthropic could afford TPUs—but some were getting cold feet.
In the end, Broadcom’s CEO held 11th-hour talks with bank executives over the weekend to eir commitments. Broadcom also agreed to higher interest rates on the loans, sweetening the deal.
With @ajsaeedy @ mattwirz for @WSJ
1
0
2
569
Every enterprise will seek to work with 3rd party partners on environment design, task creation, and human verification for their agents. I expect most successful FDE-style consultants to start promoting their own RL env design frameworks soon.
Awesome initiative @baseten and I spot some @SignalFire companies! 👀
Awesome initiative @baseten and I spot some @SignalFire companies! 👀
Today we’re announcing the Baseten Environment Network, a group of the best RL environment companies that will help our customers build custom environments for post-training.
The Network extends our applied research capacity, helping more customers build environments around and feedback.
Partners work directly with customers to create the environments, while we continue to handle model training and production deployment.
Learn more: baseten.co/blog/introduci…
0
0
16
1,938
Convergence: RL Summit is just a week away and spots are RUNNING OUT FAST.
To that end, we are going to host a 3 mile run the day before along the Embarcadero to get our systems ready. Register here, and link in comments for the main event!
luma.com/rl-rollout
To that end, we are going to host a 3 mile run the day before along the Embarcadero to get our systems ready. Register here, and link in comments for the main event!
luma.com/rl-rollout
2
1
15
820
One of the better posts I've seen actually describing what people call "RSI". Shopify has a teacher/student agent loop for continuous optimization, the question now is if you can create a similar loop to deliver net-new frontier task completion.
shopify.engineering/open-model-tas…
shopify.engineering/open-model-tas…
1
0
5
706