Anthony Platanios
@eaplatanios
Followers
646
Following
2K
Media
13
Statuses
226
VP of Research @ScaledCognition. Previously Semantic Machines (@Microsoft) and @mldcmu.
Mountain View, CA
Joined September 2017
Most enterprises we engage with face an important decision when it comes to AI in customer experience: ⛓️ Use Dialog Trees: Predictable but rigid. Customers feel like they’re in an escape room just trying to reach a human. 🤖 Use LLMs: Flexible but unreliable. Hallucinations,
0
3
6
We asked a travel bot ✈️ powered by one of those shiny LLM wrappers to extend a reservation by a couple weeks. No problem, it updated the booking and returned a (pricey but logical) total. Then we adjusted: “Actually, just extend by one day instead.” That’s when the spiral
0
3
6
Every company we talk to has the same challenge: they have existing systems that work, mostly. Dialog trees that handle millions of interactions. Business logic encoded in flow charts. They want AI but they can't afford to rip everything out and start over. It's too risky and
0
2
8
We’re actively hiring researchers! If you’re interested in building highly reliable specialized models for agentic use cases, come join us @ScaledCognition! Our work ranges from low-level modeling advances to synthetic data generation and evaluation, and is directly impacting
2
6
10
We read this paper in our reading group today. It's a cool paper; it spurred some interesting discussion. Our high-level reaction: it sure looks like the results can be explained by sampling k tokens at a time, taking the most likely sequence, then continuing.
We found a new way to get language models to reason. 🤯 No RL, no training, no verifiers, no prompting. ❌ With better sampling, base models can achieve single-shot reasoning on par with (or better than!) GRPO while avoiding its characteristic loss in generation diversity.
4
8
118
Most people don’t yet realize that systems based on general purpose LLMs are like building on jello. Models trained from the tangled mess of internet data and RL optimized for plausible sounding output are not well suited for workflow automation where precision and actual
0
7
14
The customer service world has been stuck between two extremes. On one side: rigid dialog trees. Every interaction follows predefined paths. Want to search for restaurants while booking a hotel? Sorry, that's not in the script. These systems are predictable but inflexible. On
0
6
15
Some pretty interesting research has gone into training the models behind this. It's been fun to work on this the last ~2.5 years, and I'm excited for what's coming. Also, we're hiring ;).
@ScaledCognition + @Genesys = a new era of action-driven AI for CX. Together, we’re helping enterprises deploy deterministic systems that deliver reliable, policy-aligned outcomes — built for action, not just words. Learn more →
1
2
4
Very proud of the team at @ScaledCognition and excited for this partnership with @Genesys! It's great to see our technology making an impact after months of work ranging from fundamental research to core engineering, figuring out how to bring robust agentic AI to CX.
@ScaledCognition + @Genesys = a new era of action-driven AI for CX. Together, we’re helping enterprises deploy deterministic systems that deliver reliable, policy-aligned outcomes — built for action, not just words. Learn more →
0
1
4
Will be at NAACL next week, excited to share two of our papers: FACTTRACK: Time-Aware World State Tracking in Story Outlines https://t.co/1KcL0aCWCI THOUGHTSCULPT: Reasoning with Intermediate Revision and Search https://t.co/ZGqvEeReHr Shoutout to first authors @ZhihengLyu and
0
4
10
Together with @autreche, @adishs, Krishna Gummadi, Goran Radanovic and Nina Grgić-Hlača, we have multiple open positions postdocs in AI, Computing, and Society at the MPI for Software Systems! Apply by May 13 via the new Max Planck Postdoc Program! https://t.co/IeEC4cTrDv
mpg.de
Launching in April 2025, the Max Planck Postdoc Program features structured application calls and a comprehensive support system for researchers.
1
24
80
Very excited to finally share what we’ve been working on at Scaled Cognition over the past couple of year! 🎉 Check out these videos and our website to learn more!
We’re Scaled Cognition, developing the first ever models trained specifically for agentic applications: 1. Our first system, APT-1, is now #1 on agentic benchmarks. 2. It was developed by a US team for a total cost of less than $11M. 3. Khosla Ventures led our seed round ($21M
3
2
28
Making LLMs run efficiently can feel scary, but scaling isn’t magic, it’s math! We wanted to demystify the “systems view” of LLMs and wrote a little textbook called “How To Scale Your Model” which we’re releasing today. 1/n
25
391
2K
Excited to release METAGENE-1, a 7B parameter metagenomic foundation model, built to aid in pathogen detection & pandemic monitoring. Pretrained on 1.5 trillion base pairs of DNA/RNA sequenced from wastewater. A collab w/ @USC, @PrimeIntellect, & the Nucleic Acid Observatory. 🧵
6
24
117
Excited that VPD has been selected as Oral at #CVPR2024 (90 orals in total, 0.8%). Congrats to all coauthors, and see you in Seattle! Let's distill all the powerful specialist models into one VLM! paper: https://t.co/bACXEhuzOC proj:
arxiv.org
Solving complex visual tasks such as "Who invented the musical instrument on the right?" involves a composition of skills: understanding space, recognizing instruments, and also retrieving prior...
Multimodal reasoning is hard. Even the best LMMs struggle with counting😥 Any fix for it? Introduce VPD from @GoogleAI: we teach LMMs multimodal CoT reasoning with data synthesized from LLM + vision tools, and achieve new SOTAs on many multimodal tasks!🥳 https://t.co/bACXEhuzOC
0
10
57
Calling motivated students interested in pursuing MS/PhD in ML/AI, specifically privacy & generative AI! The research group I'm starting at @iitmadras has openings! Apply by *Mar 31* directly to @DSAI_IITM or @iitmcse at https://t.co/VzV5YciG82!
Our MS/PhD Applications are now open! Apply before 🗓️31st March! #PhDposition #MS #Research - for more info on our research areas, see https://t.co/GjcIdisAcl - We have a number of exciting research centres @WSAI_IITM @cerai_iitm @IBSE_IITM @ai4bharat
0
23
65
This is really cool! First truly open-source model release where the model, the data it was trained on, and the training logs and trajectories are all released openly. Incredibly useful for the research community. Congratulations to everyone involved!
🚀 1/7 We are thrilled to launch LLM360 — pushing the frontier of open-source & transparent LLMs! Starting with Amber (7B) & CrystalCoder (7B), we are releasing brand new pre-trained LLMs with all training code, data, and up to 360 model checkpoints. 🔗 https://t.co/ZcLPtYQhdQ
0
2
12
🚀 1/7 We are thrilled to launch LLM360 — pushing the frontier of open-source & transparent LLMs! Starting with Amber (7B) & CrystalCoder (7B), we are releasing brand new pre-trained LLMs with all training code, data, and up to 360 model checkpoints. 🔗 https://t.co/ZcLPtYQhdQ
19
188
1K
Lots of people are wondering whether #GPT4 and #ChatGPT's performance has been changing over time, so Lingjiao Chen, @james_y_zou and I measured it. We found big changes including some large decreases in some problem-solving tasks: https://t.co/jgulqjvPAO
115
741
3K
Announcing FlashAttention-2! We released FlashAttention a year ago, making attn 2-4 faster and is now widely used in most LLM libraries. Recently I’ve been working on the next version: 2x faster than v1, 5-9x vs standard attn, reaching 225 TFLOPs/s training speed on A100. 1/
39
658
3K