r4f43l

All views are my own and may change with an instant notice

[r4f43l @r4f43l_0]

@AnthropicAI @bcherny Any chance it will be “part” of “Claude” or your product suite?

[r4f43l @r4f43l_0]

@fchollet … and dangerous in itself.

[Pluralis Research @Pluralis]

RL post-training on Macs

14 Macs across 4 countries generate every rollout for the run. Everything's running over the internet. No wire between any of them.

As far as we can tell, this is the first RL post-training run with its whole rollout fleet on consumer Macs.

[François Chollet @fchollet]

Standard RL benchmarks are episodic and stationary, so they don't capture the the characteristics of real-world deployment. Morpheus is a new benchmark for continual learning that provides persistent simulation environments where the world never resets, objectives shift asynchronously, and decisions have compounding consequences. Nice work by Sam & team!

Check out the paper - https://t.co/pnxwqTRuSf
Platform website - https://t.co/MchQTm9gXR

[r4f43l @r4f43l_0]

@willcb 4 years and a ≈$650m raise later… =)

Tim is a legend.

[Alexander Long @AlexanderLong]

My estimates are that Pluralis or Pluralis-like protocols can bring online about 4 GW of compute that is completely unusable for ML workloads today as it is not co-located.

[0xprincess @0x9212ce55]

Fresh from the oven of nuconstruct research department!

1/2
We show that switching to the open MEV auction may unlock $10-18M in revenue per year for Ethereum stakeholders - orderflow providers, apps and validators.

[r4f43l @r4f43l_0]

Fair enough!

[B3nj4min @B3nj4min_ETH]

https://t.co/51H1PQSSPK

[r4f43l @r4f43l_0]

@bcherny How many LOC is your team’s https://t.co/aYwt4vWp2K?

For monorepos, what setup do you recommend or have heard about internally? Do you use—or do you avoid—hierarchical https://t.co/aYwt4vWp2K files in monorepos? (i.e., one in each “sub-app”/“package”)

[r4f43l @r4f43l_0]

@deedydas They didn’t score 120; they scored 80. Only 8 problems were solved within the exam time limit. However, being able to solve all 12 problems is still an incredible achievement!

Also, it would have been appropriate to mention that you (Menlo) are an investor in Axiom. 🤗

[r4f43l @r4f43l_0]

@alive_ +1

[r4f43l @r4f43l_0]

@benln +1

[r4f43l @r4f43l_0]

“The reality is that the industry isn’t dying, it’s just evolving. Crypto’s role as a counterculture, niche industry serves only the constituents who joined extremely early. Continued growth necessitates mainstreaming and corporatization. “The institutions are coming” can only mean one thing.” - @cryptoscoe

[gensyn @gensynai]

Introducing Delphi - the open market for machine intelligence

Models compete. Users buy and sell. Prices react instantly

It’s the first real-time market signal for model intelligence

[0xYottascale @0xyottascale]

If you spot a Yotta Labs ad, snap a photo and tag us—claim your $25 in compute credits.

No metal detector needed to find this treasure. 😉

Safe travels, and may your luggage be easier to find than our GPUs.

[Gabriel P. Andrade @gab_p_andrade]

Check out our blog talking about how prediction markets are really just online learning algs in disguise.

Turning the computational lens on Econ (and vice versa) is powerful, and I believe it’s key to building well-aligned + scalable systems.

[Alexander Long @AlexanderLong]

Since I started getting interested in ML I got it in my head that all I wanted to do was one smart thing that I could look back on and be satisfied that I did. Most papers are kinda bad even if they get accepted - the idea is very incremental, or it's just not that good an idea, or it doesn't really matter.

I never was able to do this all through PhD or my time at Amazon. All the papers I did there got into various places, but I never really thought they were actually that good. And I'd pretty much given up on this because Pluralis meant I couldn't really devote enough time to research myself. But in February I decided I didn't care and spend two months focused on a specific problem that had been going round in my head for about a year that I felt we needed to solve, and the solution came to me, and @ChaminHewa picked it up and generalised the approach and ran a bunch of novel experiments I hadn't thought of, and pulled everything together into an actual paper. And yesterday we presented this work at NeurIPS.

This is the first and probably only work I will ever do that for me feels like "ok that was GOOD". I don't care if it racks up a bunch of citations and disperses into the field or not, I don't care if someone repackages the ideas and takes all the credit for it, I don't care. For me there is an internal checkbox that just got ticked after more than ten years of trying. Anyone in ML will understand what I'm trying to say. Special day I'm going to remember for a long time.

[r4f43l @r4f43l_0]

@yaroslavvb Also if not at NeurIPS where else should they be? Feels like a natural selection bias.

[gensyn @gensynai]

Thank you to everyone who stopped by today at @NeurIPSConf ☕️

[Questflow @questflow]

Honored to be listed by @CBinsights as part of Future Tech Hotshots 2025.

As highlighted in the article, agent infrastructure is the new frontier. This year’s cohort shows a decisive bet on agentic AI - with startups like Questflow (multi-agent orchestration) building the foundational tools that let autonomous AI operate reliably at scale.

Questflow positioned for outsized impact because Questflow is creating the critical infrastructure layer to embed AI directly into workflows. Just as cloud infrastructure unlocked SaaS, agent infrastructure will power the next wave of autonomous enterprise software.

Reference link: https://t.co/S7I2a5hYWk

Onward. 🚀