John Ryan (@john_pryan) profile picture

Performance and Scaling @ Isomorphic Labs

Joined August 2015
Photo by John Ryan (@john_pryan)Photo by John Ryan (@john_pryan)Photo by John Ryan (@john_pryan)Photo by John Ryan (@john_pryan)Photo by John Ryan (@john_pryan)Photo by John Ryan (@john_pryan)Photo by John Ryan (@john_pryan)Photo by John Ryan (@john_pryan)Photo by John Ryan (@john_pryan)Photo by John Ryan (@john_pryan)
John Ryan retweeted
person who thinks AI will prove the Riemann hypothesis in 2026 but that Apple won’t deliver an LLM-powered Siri until 2027
183
525
198
6,443
572,784
John Ryan retweeted
Hi all, I have a new result I'm excited to share. It turns out that *all existing implementations* of softmax and cross-entropy (that's JAX, PyTorch, textbooks, etc) have a subtle bug --- two bugs, actually. A simple rewrite fixes both, and eliminates "slingshots" when grokking:
133
796
156
6,579
888,778
John Ryan retweeted
Introducing AlphaProtein Novo, a pipeline for de novo enzyme design! In collaboration with @francesarnold's group, we developed AP Novo and used it to get SoTA "novel-scaffold" activity on 2 benchmark reactions and create enzymes to synthesize the pharmaceutical motif piperidine and degrade the environmental toxin DEHP. Work led by @ZvxyWu @ZiqiLi0513 with teammates @alexechu_ @kelly_ruijie @schulluc @jacobjinkelly @rosaliags @davidla @solserpens @j_reisenbauer, @pengliu_group, Josh, Thomas, Amy, Tristan, Harshnira, and others. It was a joy and privilege to work with you all! Preprints and code: biorxiv.org/content/10.64898… biorxiv.org/content/10.64898… github.com/google-deepmind/a…
20
179
24
1,024
139,822
John Ryan retweeted
Ireland has ended up a vetocracy.
126
74
35
693
194,540
John Ryan retweeted
dream harder
94
300
31
2,500
276,730
There are still multiple paths to make this happen. They all involve nuking two common delusions in outer space: 1. We can win at the application layer 2. Let’s give up on LLM and do 15m LeWorldModel.
Not having any EU competitive labs (plural intended) is about to get very painful.
12
13
3
148
12,157
John Ryan retweeted
people aren’t celebrating this pancreatic cancer breakthrough enough. it should be international news
One of the most amazing things I’ve ever seen: a standing ovation for the full Daraxonrasib results I feel inspired and energised, to put it mildly — we have a targeted therapy for pancreatic cancer now, and nothing is undruggable anymore
87
1,476
33
11,872
811,356
I’ve always believed the No.1 application of AI should be to improve human health. That work started with AlphaFold, and now at @IsomorphicLabs with the mission to reimagine drug discovery and one day solve all disease! We are turbocharging that goal with $2.1B in new funding.
703
2,599
443
20,969
3,191,531
VPS 2.0 Outperforms GPS Alone. Niantic Spatial’s Visual Positioning System (VPS) matches live camera data to a spatial model for precise 6DoF localization—indoors, outdoors, and in complex environments where GPS fails. 🔹 ~10-15cm accuracy in mapped areas 🔹 Works in GPS-denied environments 🔹 Persistent spatial anchors Explore this deep dive to learn how VPS works: hubs.ly/Q04csZsd0 #NianticSpatial #Scaniverse #VPS #GIS #GeospatialAI #GEOINT #PhysicalAI #AI #Robotics
4
32
1
194
19,917
John Ryan retweeted
Here at [famous scientist’s name].ai, we’re developing tools to accelerate science. Unlike academia, which has stifled the production of high quality scientific work by demanding it be correct, here at [fsn].ai, we know that you can just do (wrong) things.
36
224
7
2,464
79,312
John Ryan retweeted
So we're slowly getting more information on what spooked Google into setting their post-quantum migration timeline to 2029. Two puzzle pieces dropped today.
10
66
4
1,155
272,970
John Ryan retweeted
We asked every version of Claude to make a clone of Claude(dot)ai, including today’s Sonnet 4.5… see what happened in the video
122
273
118
3,832
458,451
John Ryan retweeted
Maybe to one's surprise, taking KL estimates as `kl_loss` to minimize does *not* enforce the KL. This implementation, however, is quite common in open source RL repos and recent research papers. In short: grad of an unbiased KL estimate is not an unbiased estimate of KL grad.
14
55
3
654
71,533
Meet MapAnything – a transformer that directly regresses factored metric 3D scene geometry (from images, calibration, poses, or depth) in an end-to-end way. No pipelines, no extra stages. Just 3D geometry & cameras, straight from any type of input, delivering new state-of-the-art results 🚀 One universal model enables SoTA for: 🔥 Mono Depth Estimation 🔥 Multi-View SfM 🔥 Multi-View Stereo 🔥 Depth Completion 🔥 Registration … and many more possibilities! – plus everything is metric 🎯 We release code for data processing, training, benchmarking & ablations – everything Apache 2.0! Details & Links 👇
29
130
31
732
123,917
John Ryan retweeted
Arc Institute trained their foundation model Evo 2 on DNA from all domains of life. What has it learned about the natural world? Our new research finds that it represents the tree of life, spanning thousands of species, as a curved manifold in its neuronal activations. (1/8)
10
53
14
365
78,254
My team @IsomorphicLabs is hiring ML Research Scientist to work on the next generation of interaction models for drug design. We are looking for a solid research track record, ambition to revolutionize drug discovery. Please reach out to @benrozemberczki bit.ly/45nhyg2
5
33
4,041
🚀 After two+ years of intense research, we’re thrilled to introduce Skala — a scalable deep learning density functional that hits chemical accuracy on atomization energies and matches hybrid-level accuracy on main group chemistry — all at the cost of semi-local DFT. ⚛️🔥🧪🧬
5
62
14
288
33,446
We are excited to announce the expansion of our collaboration with @Novartis, doubling the number of research programs just over a year after initiating our partnership. This expansion is testament to the great progress we’ve made together and further unlocks the potential to ultimately advance human health. Read more here: bit.ly/4i669WZ #AI #drugdiscovery #machinelearning
10
56
11
298
50,590
Speculative sampling accelerates inference in LLMs by drafting future tokens which are verified in parallel. With @ValentinDeBort1 , @agalashov & @ArthurGretton, we extend this approach to (continuous-space) diffusion models: arxiv.org/abs/2501.05370
6
51
3
251
18,334