Vindula Jayawardana

Research Scientist @ToyotaResearch | PhD @MIT | Prev @Nvidia @Toyota @wso2

[Vindula Jayawardana @vindulajay]

Our work is in the spotlight today on MIT front page. #ClimateAction #ArtificialIntelligence #MIT #ReinforcementLearning

[Massachusetts Institute of Technology (MIT) @MIT]

No one likes sitting at a red light, or wasting fuel and producing emissions while idling. What if motorists could pace themselves to arrive when the light is green? A LIDS team’s new machine-learning approach could keep traffic flowing smoothly. https://t.co/LLIk1iVA5b

[Vindula Jayawardana @vindulajay]

Creative idea!

[Sasha Malysheva @aimalysheva]

meet @mostik_ai!

what happens when you put 12 PhDs in one room for four months? first place on the ARC-AGI leaderboard, which I can't say much about while the competition is still running. and this, which I can.

everyone's arguing about whether open models will catch up to frontier models. we think it's the wrong question. here's the one we pose: why does a frontier model have to generate your answer at all, when the only thing you need from it is the reasoning?

we do this by enabling models to communicate in latent space. through our protocol, hidden states pass straight from a frontier model into a small one running on your infrastructure -- no text between them, and neither model is fine-tuned. two models from different families, sharing reasoning, both left untouched.

how do we know it works? we tested it on a setup where a 753B model reads the problem, and a 4B edge-class model writes the answer. with this approach, we get results 80% as accurate as the frontier model, but at 20x faster performance.

we're committed to preventing frontier model lock-in and are already partnering with inference providers to accelerate open-weight adoption. we've done this between 15 of us, in four months, 12 PhDs and a Fields medalist, backed by @generalcatalyst

WIRED has the first external account of the company and the work: https://t.co/tP8nItCsDl
full writeup, the setup, and all the numbers: https://t.co/C9NZ5vtV1V

[Vindula Jayawardana @vindulajay]

Was great to collab on this cool project! Check it out!

[Surabhi S Nath @surabhisnath]

🚨Paper‼️Alert 🔈

Work from my lovely internship 🚢 at @ToyotaResearch is out 💃🏻

A thread 🧵⬇️

https://t.co/htUEwUBVwc

[Vindula Jayawardana @vindulajay]

Tabracadabra!! Hit us up with your feedback 😀

[Omar Shaikh @oshaikh13]

We upgraded Tabracadabra 🎉 to bring an entire context-aware assistant (not just tab to autocomplete!) to any textbox. It's pretty great if you hate switching between the chat interface and what you're working on. We're also open-sourcing, so you can try it out!🧵

[Vindula Jayawardana @vindulajay]

RT @Diyi_Yang: Cool to see Chronicle from @OpenAI 👀This is closely related to what we explored in GUM https://t.co/zwyCjy1FJj. Great to see…

[Association for Computing Machinery @TheOfficialACM]

Meet the recipients of the 2024 ACM A.M. Turing Award, Andrew G. Barto and Richard S. Sutton! They are recognized for developing the conceptual and algorithmic foundations of reinforcement learning. Please join us in congratulating the two recipients! https://t.co/GrDfgzW1fL

[Cathy Wu (Hiring PhD/postdocs) @wucathy]

👼 As an applied RL researcher, this is the most optimistic I have been about RL in years. It feels like seeing the light at the end of the tunnel when RL training starts working reliably. Without a ton of compute or tuning. Very excited for what is to come. Here is what we did👇

[Vindula Jayawardana @vindulajay]

Our #NeurIPS24 paper on model-based transfer learning for contextual reinforcement learning is featured in MIT News today. Check it out here: https://t.co/50siMx29q2

#neurips24 #reinforcementlearning #autonomousvehicles

[Cathy Wu (Hiring PhD/postdocs) @wucathy]

Reinforcement learning (RL) is surprisingly brittle to contextual variations in tasks. Our new method (NeurIPS 2024) for solving contextual RL problems achieves 5-50x better sample efficiency on standard & traffic benchmarks. Featured today by MIT news! https://t.co/dR1xK5tgoT

[Vindula Jayawardana @vindulajay]

@EugeneVinitsky Same situation here. High scores but got rejected. At least my AC hinted that they had to reject to meet the 25%.

[Vindula Jayawardana @vindulajay]

@ramith__ @ChamiraEduss @prathapakd @rangarodrigo @nawodya @sokrypton Congratulations Ramith! All the best!

[Jeremy Hsu (@jeremyhsu.bsky.social) @jeremyhsu]

A simple driving choice when approaching traffic intersections could save a significant amount of carbon emissions (and fuel).

This can be programmed into driverless vehicles or cars with adaptive cruise control - but drivers can also manually do it!

https://t.co/vASgNJ1qkU

[Vindula Jayawardana @vindulajay]

@newscientist @jeremyhsu @wucathy Direct article link

https://t.co/qI3DqFSKuW

[Vindula Jayawardana @vindulajay]

✨Our latest research on AI-driven eco-driving has been featured in @newscientist , in a well-written piece by @jeremyhsu. This is joint work with my advisor @wucathy and the #Greenwave team.

#autonomousvehicles #reinforcementlearning #traffic #machinelearning #selfdrivingcars

[New Scientist @newscientist]

Like many of us, you may be speeding up to intersections then slamming the brakes – but you can help greatly reduce carbon emissions by gliding up slowly
https://t.co/3YgmIIJfyv

[Vindula Jayawardana @vindulajay]

@tim_garipov @samikaski @phillip_isola Congratulations!! @tim_garipov

[Eugene Vinitsky 🦋 @EugeneVinitsky]

We deployed 100 deep RL cruise controllers into rush-hour highway traffic to smooth traffic flow and reduce everyone’s energy consumption. Our AVs were decentralized and used standard radar, making our controllers deployable on most cars
1/n

[Vindula Jayawardana @vindulajay]

@Toyota @ieee_ras_icra @SiruiLi01 @wucathy @MIT #reinforcementlearning #robotics #selfdriving #autonomousvehicles #ecodriving

[Vindula Jayawardana @vindulajay]

I will be presenting my summer internship research with @Toyota at @ieee_ras_icra 2024. Grateful to my hosts Yashar Farid and Kentaro Oguchi at @Toyota and my collaborators @SiruiLi01 and @wucathy at @MIT.

[Vindula Jayawardana @vindulajay]

@Toyota @ieee_ras_icra @SiruiLi01 @wucathy @MIT In this work, we present a new learning framework to learn vehicle control policies that generalize across problem variations. For more details, check out our paper and the website below.

Paper: https://t.co/OaUuXszjew
Website: https://t.co/X62mMswads

[Vindula Jayawardana @vindulajay]

I am co-organizing the Autonomous Vehicles Across Scales workshop at RSS 2024 with Cameron Hickert @wucathy @amans1nha @MashaItkina and Nikita Jaipuria. Consider submitting your work.

Website: https://t.co/eJivxPKiGw

#rss2024 #autonomy #autonomousvehicles #selfdriving #robotics

[Cathy Wu (Hiring PhD/postdocs) @wucathy]

The 𝐀𝐮𝐭𝐨𝐧𝐨𝐦𝐨𝐮𝐬 𝐕𝐞𝐡𝐢𝐜𝐥𝐞𝐬 𝐀𝐜𝐫𝐨𝐬𝐬 𝐒𝐜𝐚𝐥𝐞𝐬 (𝐀𝐕𝐀𝐒) 𝐖𝐨𝐫𝐤𝐬𝐡𝐨𝐩 at RSS 2024 is seeking submissions! We will bring together researchers across the three ‘scales’ of AV research – ego vehicle, traffic, & network. Deadline is May 24th (4-8 pages).

[MIT CEE @MIT_CEE]

Join the next CEE Graduate Seminar Series with speakers Vindula Jayawardana and Dat Quoc Ha 4/19 at 4PM in 48-316.

[Vindula Jayawardana @vindulajay]

@nandanamihindu Felt here in Boston, too!