Arnie Ramesh's banner

Arnie Ramesh

@arnie_hacker • 6,143 subscribers

CS grad @ ETH Zürich | prev @shipfr8 @apple @aws | angel

Videos

Anya Rossi

sweetdream.ai

SweetDream.ai•Sponsored•Livecam

Watch Anya Live

Anya is streaming live right now! Join her private show and enjoy exclusive content.

Exclusive private shows

1.2k viewers online

Private Show

Join now for exclusive access

Free preview available • Premium content

CounterStrike-1K is almost here. > 1K hours of multi-pov, pro gameplay > Video (720p@32fps, h.264) + Audio > action-annotated + rich game metadata > Optimized for training (GOP=32+WDS) Cleaning the data now - whoever flags the most render bugs gets 50$

CounterStrike-1K is almost here. > 1K hours of multi-pov, pro gameplay > Video (720p@32fps, h.264) + Audio > action-annotated + rich game metadata > Optimized for training (GOP=32+WDS) Cleaning the data now - whoever flags the most render bugs gets 50$

162,230 просмотров • 1 месяц назад

Some updates on the CS:GO 100K hours dataset :) So far: > Idempotent, fault-tolerant (w/ retries) CS2 gameplay rendering on Amazon Web Services infra (EC2, SQS, DDB, S3) > Video(720p@30fps) + Audio + *GT Keyboard/Mouse > End-to-end benchmarking script w/ in-depth CloudWatch logging

Some updates on the CS:GO 100K hours dataset :) So far: > Idempotent, fault-tolerant (w/ retries) CS2 gameplay rendering on Amazon Web Services infra (EC2, SQS, DDB, S3) > Video(720p@30fps) + Audio + *GT Keyboard/Mouse > End-to-end benchmarking script w/ in-depth CloudWatch logging

58,077 просмотров • 2 месяцев назад

cs2 rendering finally works :) - patched cs demo manager spec_player bug - multi-cs2 instances per worker - multi-workers (scalable, fault-tolerant) what should i generate? - 100hrs of dust2 gameplay (synchronized, 10-player w/ audio)? - 100hrs, but of 7 maps (same set-up)

cs2 rendering finally works :) - patched cs demo manager spec_player bug - multi-cs2 instances per worker - multi-workers (scalable, fault-tolerant) what should i generate? - 100hrs of dust2 gameplay (synchronized, 10-player w/ audio)? - 100hrs, but of 7 maps (same set-up)

38,613 просмотров • 1 месяц назад

The future is being built in Europe. Zurich Robotics Hack🇨🇭

The future is being built in Europe. Zurich Robotics Hack🇨🇭

138,900 просмотров • 1 год назад

working towards action-conditioned video diffusion models for now this is just cs2 gameplay, and i parse the keypresses from the .dem game file next steps will be to work on a scalable data loader then coding the model

working towards action-conditioned video diffusion models for now this is just cs2 gameplay, and i parse the keypresses from the .dem game file next steps will be to work on a scalable data loader then coding the model

31,090 просмотров • 3 месяцев назад

Zurich Robotics Hack🇨🇭 If you were in SF you missed out fr

Zurich Robotics Hack🇨🇭 If you were in SF you missed out fr

56,687 просмотров • 1 год назад

In just one week, Binh Pham and I trained a full-body Unitree G1. Here's a recap: 1. Secured a Unitree G1 humanoid through a LinkedIn post 2. Deployed TWIST2 full-body teleoperation pipelines 3. Adapted TWIST2 for Zed stereo camera & collected full-body teleoperation samples (carried by Binh Pham ) 4. Adapted & fine-tuned NVIDIA Gr00T N1.5 VLA on the TWIST2 public datasets, which I fine-tuned on an 8xNVIDIA H100 Cluster. We picked Gr00T N1.5 as it was trained with Unitree G1 embodiment data. 5. Adapted the TWIST2 codebase to stream in the actions from Gr00T via ZMQ using a co-located NVIDIA H100 for ~200ms inference latency 6. Tested the model in sim, then deployed to the real-world Unitree G1. We streamed a training sample observation to the VLA (as we didn't want to break robot in case real observations were OOD) We were the first team in the world to deploy the full TWIST2 data collection pipeline to the unitree g1 :) Much more work ahead though, which I'll work on as a side-project over the next months: 1. Exploring the various types of 'world models': video backbones, dynamics models, v-jepa-2 models. I believe these will generalize better & train much more data-efficiently than VLM backbones 2. Speeding up inference - I believe low-latency robotics inference will be a big challenge. There are many works in video diffusion which I'd like to test (e.g. SageAttention, SparseAttention, Drifting Models). Perhaps also writing custom CUDA kernels. 3. Economics of inference scaling :) What will be the compute demands as we scale inference up to millions of humanoids? Will it run on edge or on distributed 'co-located' inference clusters? These are questions I'd like to answer. Adapted TWIST2 codebase: Adapted Gr00T-N1.5 codebase: The ETH Robotics Club are doing a cool GTC Golden ticket competition with NVIDIA , so this is my submission :) The DGX Spark compute will get me a long way with initial prototyping & especially working on inference optimization for next-gen Blackwell GPUs #NVIDIAGTC #GOLDENTICKET #ETHRC

In just one week, Binh Pham and I trained a full-body Unitree G1. Here's a recap: 1. Secured a Unitree G1 humanoid through a LinkedIn post 2. Deployed TWIST2 full-body teleoperation pipelines 3. Adapted TWIST2 for Zed stereo camera & collected full-body teleoperation samples (carried by Binh Pham ) 4. Adapted & fine-tuned NVIDIA Gr00T N1.5 VLA on the TWIST2 public datasets, which I fine-tuned on an 8xNVIDIA H100 Cluster. We picked Gr00T N1.5 as it was trained with Unitree G1 embodiment data. 5. Adapted the TWIST2 codebase to stream in the actions from Gr00T via ZMQ using a co-located NVIDIA H100 for ~200ms inference latency 6. Tested the model in sim, then deployed to the real-world Unitree G1. We streamed a training sample observation to the VLA (as we didn't want to break robot in case real observations were OOD) We were the first team in the world to deploy the full TWIST2 data collection pipeline to the unitree g1 :) Much more work ahead though, which I'll work on as a side-project over the next months: 1. Exploring the various types of 'world models': video backbones, dynamics models, v-jepa-2 models. I believe these will generalize better & train much more data-efficiently than VLM backbones 2. Speeding up inference - I believe low-latency robotics inference will be a big challenge. There are many works in video diffusion which I'd like to test (e.g. SageAttention, SparseAttention, Drifting Models). Perhaps also writing custom CUDA kernels. 3. Economics of inference scaling :) What will be the compute demands as we scale inference up to millions of humanoids? Will it run on edge or on distributed 'co-located' inference clusters? These are questions I'd like to answer. Adapted TWIST2 codebase: Adapted Gr00T-N1.5 codebase: The ETH Robotics Club are doing a cool GTC Golden ticket competition with NVIDIA , so this is my submission :) The DGX Spark compute will get me a long way with initial prototyping & especially working on inference optimization for next-gen Blackwell GPUs #NVIDIAGTC #GOLDENTICKET #ETHRC

14,815 просмотров • 3 месяцев назад

Больше нет контента для загрузки