Ilia Larchenko PRO

IliaLarchenko

AI & ML interests

I am a Data Science Director with a diverse technical and business background. I live in Bangkok and work in Agoda, where I lead multiple DS and ML teams. I was an active Kaggler in TOP-20 of a global competition ranking, Competitions and Notebooks Master

Recent Activity

posted an update about 3 hours ago

I am presenting Decoder-Only Transformer (DOT) Policy a simple Behavioral Control policy that outperforms SOTA models on two simple benchmark tasks: ✅ PushT (pushing an object to a goal) – 84% success on keypoints, 74% on images (previous best: 75% / 69%) ✅ ALOHA Insert (precise bimanual insertion) – 30% success (previous best: ~21%) The best part? DOT is much smaller (sometimes 100 times less parameters) than previous SOTA models, trains faster, and avoids complexity: 🚫 No generative models (Diffusion, VAE, GANs) 🚫 No discretization/tokenization of actions 🚫 No reinforcement learning or multi-stage training ✅ Just learns from human demos, plain and simple This is still early — more complex real-life tasks need testing, and no guarantees it will actually work well there, but I think it's interesting to share. Sometimes, simpler approaches can be just as effective (or even better) than complex ones. 🔗 Open-source code and detailed description: https://github.com/IliaLarchenko/dot_policy Trained models on Hugging Face: https://huggingface.co/IliaLarchenko/dot_pusht_keypoints https://huggingface.co/IliaLarchenko/dot_pusht_images https://huggingface.co/IliaLarchenko/dot_bimanual_insert

new activity 1 day ago

lerobot/xarm_lift_medium_image:Incorrect dataset

updated a model 4 days ago

IliaLarchenko/dot_pusht_keypoints_best

View all activity

Organizations

Posts 1

Post

210

I am presenting Decoder-Only Transformer (DOT) Policy a simple Behavioral Control policy that outperforms SOTA models on two simple benchmark tasks:

✅ PushT (pushing an object to a goal) – 84% success on keypoints, 74% on images (previous best: 75% / 69%)
✅ ALOHA Insert (precise bimanual insertion) – 30% success (previous best: ~21%)

The best part? DOT is much smaller (sometimes 100 times less parameters) than previous SOTA models, trains faster, and avoids complexity:
🚫 No generative models (Diffusion, VAE, GANs)
🚫 No discretization/tokenization of actions
🚫 No reinforcement learning or multi-stage training
✅ Just learns from human demos, plain and simple

This is still early — more complex real-life tasks need testing, and no guarantees it will actually work well there, but I think it's interesting to share. Sometimes, simpler approaches can be just as effective (or even better) than complex ones.

🔗 Open-source code and detailed description: https://github.com/IliaLarchenko/dot_policy

Trained models on Hugging Face:
IliaLarchenko/dot_pusht_keypoints
IliaLarchenko/dot_pusht_images
IliaLarchenko/dot_bimanual_insert