Angehefteter Tweet
Modal
1.5K posts
Modal
@modal
AI infrastructure that developers love 💚 Run inference, sandboxes, batch processing, training, and many other things on Modal
New York City Beigetreten Temmuz 2022
151 Folgt30.9K Follower
Modal retweetet
Modal retweetet
Modal retweetet
Last fall, we shared our deep dive on FA4 internals.
But we didn't stop at grokking the kernel.
Since then, we've been developing improvements for inference performance and upstreaming them.
This blog post explains those contributions.
modal.com/blog/flash-att…
English
i joined @modal :)
we're creating a publication
The Modal Review - a literary print and digital publication, with a balanced mixture of dispatches, guest essays, poetry and art
English
Modal retweetet
Tried to squeeze the most important bits about the entire stack for cloud deployment of transformer inference, from application layer concerns to hardware, debugging, and o11y, into one talk. Had to operate at a very high tok/s!
youtube.com/watch?v=ZUdIsR…
YouTube
English
The future of training is open source.
Super excited to announce that we've joined forces with HuggingFace, Nvidia, Meta, Mercor and other leading companies to support OpenEnv :)
Ben Burtenshaw@ben_burtenshaw
So excited to be opening up OpenEnv to the whole community. It will now be owned by @huggingface , Meta-PyTorch, @reflection_ai , @UnslothAI , @modal, @PrimeIntellect , @NVIDIAAI , @mercor_ai , and @fleet_ai . the reason is: frontier labs train the model and the harness together, so the model is fitted to its harness. that coupling is a chunk of why claude code and codex feel so good. open source can't do that. you bring whatever harness, whatever model, whatever env, whatever trainer. which is the whole point of open source and also the problem for training. openenv is the socket in between all of this. in short: it's a protocol layer, not a reward framework. it does not have opinions about your rewards or your training loop. those live in the libs that are actually good at them. read more in the blog post. it's early, come break it.
English
Two big updates on my end:
1. I’m moving on from @PrefectIO after three wonderful years.
2. I started this week at @modal as Member of Product Staff.
I’m so incredibly grateful to @jlowin , Chris White, Thomas Egbert and the team at Prefect. I got to build world class open source software and be a part of Prefect and @fastmcp's vertical growth.
Three years ago I joined Modal as a user. It was March 2023 and I was two days into a losing fight with ECS. @PhysicsBliss told me to check out Modal. @bernhardsson replied to my signup email in two minutes; by that night, days of work had collapsed into hours. Modal sparked joy and I was preaching the gospel to anyone who would listen.
That didn’t stop when I joined Prefect. It’s rare to find kindred spirits, who share your scar tissue, who think about software and developer experience the way you do. Unsurprisingly then, Modal was my default recommendation to customers and I kept working with Erik, @akshat_b and @charles_irl to build integrations and do events together.
So I'm making it official. I get to do full-time what I've been doing as a fan for three years: help people build without friction and turn months of work into hours.
I couldn't be more excited to build alongside Erik, Akshat, Charles, and the whole team. Here's to the next chapter.
English
With today's launch of Nemotron 3 Ultra, @nvidia continues to expand its investment in open-source AI. Their flagship frontier-reasoning model, built for long-running autonomous agents, is available Day 0 on Modal.
- 550B with 55B active parameters
- Hybrid Transformer-Mamba MoE architecture
- 1M context
- Up to 5x faster inference
- Up to 30% lower cost
NVIDIA@nvidia
Introducing NVIDIA Nemotron 3 Ultra. A frontier smart open model built for long-running agents that need to plan, reason, use tools and keep working across complex coding, research and enterprise workflows. Up to 5x faster inference and up to 30% lower cost for agentic tasks. Learn more: nvda.ws/4x9nGps
English
Modal retweetet
we're hosting some parties to celebrate our C 💚
exclusive swag at both ofc
Modal@modal
We're bringing together our friends and community to celebrate our Series C. Join us at Noguchi's Sunken Garden in NYC on June 16th or at the Legion of Honor in SF on June 25th. Invites are limited, apply here: modal.com/c-function
English
Modal retweetet
I love to see @modal being used for biology and cutting-edge research like this. Very cool work from the team at @biohub to push open models forward in protein design and comp bio.
Here's how to run it on Modal: modal.com/docs/examples/…
Thomas Hayes@THayes427
I’m so excited about the launch of ESMFold2, ESMC, and the new ESM Atlas. This was a massive team effort, and I’m grateful to have worked with such an incredible group @biohub. A headline result I’m especially excited about: ESMFold2 can design minibinders and antibodies with nanomolar affinity, target selectivity, and functional activity against therapeutically relevant targets. Today, we’re sharing the full binder design protocol.
English
We're bringing together our friends and community to celebrate our Series C.
Join us at Noguchi's Sunken Garden in NYC on June 16th or at the Legion of Honor in SF on June 25th.
Invites are limited, apply here: modal.com/c-function
English
very exciting work and thrilled to be working on RL this summer at @modal!
Modal@modal
Reinforcement learning has exploded on Modal, and we've been cooking. Here's a review of lessons learned helping teams train at scale, the patterns we kept seeing, and an open-source library to get started with RL on Modal quickly.
English
personal update: i'm joining @modal
modal unequivocally is the best AI infra company.
i've been a customer for about a year now.
i'm in london for two months, if any startup seed and above is willing and able to move to modal. DM me. I will help out personally
English
Modal retweetet
At @modal, we're working to make sure OSS RL frameworks have all the techniques necessary to train frontier open-weights models.
Delta compression is key, but the job's not done. There are still lots of open problems around weight sync, auto-scaling, & cross-cluster training.
My DMs are open!
slime@slime_framework
@FireworksAI_HQ + @cursor_ai highlighted why delta-compressed weight sync matters for RL at frontier scale. slime brings this capability to OSS: lossless delta sync for Megatron ↔ SGLang disaggregation — ship deltas, not full checkpoints. This is another step toward a fully open-source stack where rollout/inference and training are truly decoupled and deployed separately. PR: github.com/THUDM/slime/pu…
English
Cyber attackers don't wait for you to spin up infrastructure.
How @DoppelHQ uses Modal's elastic compute to scale inference, cut training overhead, and parallelize experimentation 👇
Doppel@DoppelHQ
How does Doppel stay ahead of phishing campaigns that spin up millions of domains at once? Machine learning engineers William Gill and Ishana Shastri explain how they use @modal to: - React to sudden cyberattack influxes with highly elastic GPU clusters. - Replace complex Cloud Run and Docker setups with simple decorators in a single code file. - Make experimentation cost-efficient by running parallel experiments in the background. Learn more: doppel.com/blog/how-doppe…
English