Video wird geladen...

Video konnte nicht geladen werden

Beim Laden dieses Videos ist ein Problem aufgetreten. Dies könnte an einem vorübergehenden Netzwerkproblem liegen oder das Video ist möglicherweise nicht verfügbar.

in the lectures below, i hold your hand through low-level LLM systems engineering. it includes everything up to TODAY! 1) pytorch tensors 2) large matmul on cpu vs gpu 3) JAX (and why xAI uses it instead of pytorch) 4) raw cuda kernels and global threading indexing 5) triton... design philosophy and softmax example 6) HIP kernels 7) mapping out the ENTIRE ecosystem + differences between CUDA and ROCm/HIP (BLAS, FFT, DNN) 8) cutlass and cute-dsl 9) pretraining, finetuning, rl, unsloth, axolotl, megatron-lm, deepspeed, nanogpt, nanochat 10) training vs inference, inference serving problems, throughput vs latency vs concurrency scaling, vllm, sglang, tensorrt-llm, tensorrt, llama.cpp, exllamav2, exllamav3, benchmark comparisons 11) projects/companies using llms to generate SOTA cuda/triton kernels 12) luminal inference 13) mojo/modular/maxshow more

Elliot Arledge

44,604 subscribers

58,522 Aufrufe • vor 1 Jahr •via X (Twitter)

Wissenschaft & Technologie Bildung

Anya Rossi• Live Now

Private livecam show

47 Kommentare

Profilbild von Andy

Andyvor 1 Jahr

cool, do you have any good resources for optimizing inference on cpus ?

Profilbild von Elliot Arledge

Elliot Arledgevor 1 Jahr

@grok show him llama.cpp lore

Profilbild von Grok

Grokvor 1 Jahr

Llama.cpp is a C/C++ library for efficient LLM inference, started by Georgi Gerganov in 2023 to run models like Llama on everyday hardware. Its "lore" includes a focus on "inference at the edge" – enabling fast, local AI without cloud dependency. Key for CPU optimization: quantization (down to 1.5-bit) reduces model size/memory, AVX/AVX2/AVX512 acceleration, and BLAS backends boost speed. Check the Changelog podcast for dev backstory. Great for low-resource setups!

Profilbild von The AI Guy

The AI Guyvor 1 Jahr

Can you upload it to YouTube

Profilbild von Elliot Arledge

Elliot Arledgevor 1 Jahr

yep

Profilbild von Daniel

Danielvor 1 Jahr

what software do you use to record your screen? I am asking for the zoom effect

Profilbild von Elliot Arledge

Elliot Arledgevor 1 Jahr

screenstudio

Profilbild von Daniel

Danielvor 1 Jahr

thanks man

Profilbild von vineet

vineetvor 1 Jahr

Put it on YT bro, who watches vids on x

Profilbild von Shini

Shinivor 1 Jahr

Lecture looking fire. Had to add it to weekend watch list.

Profilbild von sharpeye

sharpeyevor 1 Jahr

God’s content

Profilbild von paneerchilli65

paneerchilli65vor 1 Jahr

weekend watch 🫡

Profilbild von subho ghosh

subho ghoshvor 1 Jahr

way to go!

Profilbild von Elliot Arledge

Elliot Arledgevor 1 Jahr

thx

Profilbild von Siolu

Sioluvor 1 Jahr

oh look here, the cook is back to cooking a feast

Profilbild von Elliot Arledge

Elliot Arledgevor 1 Jahr

i am

Profilbild von Maxen

Maxenvor 1 Jahr

Nice, I'll definitely watch. Is there a link on Youtube?

Profilbild von Elliot Arledge

Elliot Arledgevor 1 Jahr

uploading...

Profilbild von Rohan Sharma

Rohan Sharmavor 1 Jahr

fire 🔥

Profilbild von Elliot Arledge

Elliot Arledgevor 1 Jahr

i try

Profilbild von Rajneel

Rajneelvor 1 Jahr

You drop exactly what I needed when I needed before cuda and now this thxxx

Profilbild von Elliot Arledge

Elliot Arledgevor 1 Jahr

cool

Profilbild von ed

edvor 1 Jahr

of course. don't let me sleep man

Profilbild von Elliot Arledge

Elliot Arledgevor 1 Jahr

lol

Profilbild von CrazyAI Tech

CrazyAI Techvor 1 Jahr

Really almost everything! Youtube subscribed!

Profilbild von mateo

mateovor 1 Jahr

jesus elliot

Profilbild von Elliot Arledge

Elliot Arledgevor 1 Jahr

daddy is very much back

Profilbild von mateo

mateovor 1 Jahr

daddy never left

Profilbild von Nafiz

Nafizvor 1 Jahr

YT link?

Profilbild von Elliot Arledge

Elliot Arledgevor 1 Jahr

on it

Profilbild von gunwook fan

gunwook fanvor 1 Jahr

Is this beginner friendly

Profilbild von Elliot Arledge

Elliot Arledgevor 1 Jahr

yes for the most part

Profilbild von 0xmaddy | Techie

0xmaddy | Techievor 1 Jahr

Essential curriculum—understanding CUDA/Triton fundamentals separates real ML engineers from API consumers. Low-level systems knowledge compounds your career trajectory. Ship it!

Profilbild von Yousef | Developer | e/acc

Yousef | Developer | e/accvor 1 Jahr

YT link please 🙏

Profilbild von Elliot Arledge

Elliot Arledgevor 1 Jahr

on it

Profilbild von mohit

mohitvor 1 Jahr

insane

Profilbild von Elliot Arledge

Elliot Arledgevor 1 Jahr

thx

Profilbild von maharshi

maharshivor 1 Jahr

the amount of effort you put behind this is insane, so cool elliot

Profilbild von Leah

Leahvor 1 Jahr

I wanna hold hands (professionally)

Profilbild von Mohit

Mohitvor 1 Jahr

create episodes that would be useful

Profilbild von Anay

Anayvor 1 Jahr

Weekend is sorted!

Profilbild von Ex_Machina

Ex_Machinavor 1 Jahr

Love you

Profilbild von sohail

sohailvor 1 Jahr

lovely running through this so far also love that I can see the creatine + cuda in this (w the overears + pump cover - makes it feel like an authentic walkthrough lmao) next we need a ppl vs arnold vs upper lower breakdown

Profilbild von Mahid from Openrobot

Mahid from Openrobotvor 1 Jahr

legend

Profilbild von maxwell

maxwellvor 5 Monaten

Thanks

Profilbild von manv

manvvor 11 Monaten

Thank you...

Profilbild von Ananth Narayan

Ananth Narayanvor 11 Monaten

Wth this is amazing!

Ähnliche Videos

CMU PhD who built the kernels NVIDIA now ships in TensorRT-LLM explained fast attention in 68 minutes - better than $1200 GPU programming courses. pick the attention pattern -> generate a fused CUDA kernel -> drop it into vLLM/SGLang -> same GPU, way more tokens per second. That loop is why FlashInfer now powers inference at NVIDIA, vLLM, SGLang, and half the serving stacks you use. FlashInfer + Triton + JIT-compiled kernels + paged-KV attention - that's the stack.

CMU PhD who built the kernels NVIDIA now ships in TensorRT-LLM explained fast attention in 68 minutes - better than $1200 GPU programming courses. pick the attention pattern -> generate a fused CUDA kernel -> drop it into vLLM/SGLang -> same GPU, way more tokens per second. That loop is why FlashInfer now powers inference at NVIDIA, vLLM, SGLang, and half the serving stacks you use. FlashInfer + Triton + JIT-compiled kernels + paged-KV attention - that's the stack.

h100envy

32,605 Aufrufe • vor 4 Monaten

Understanding GPUs will change your AI Engineering journey beyond agents and LLM. Rise of LLMs and deep learning introduced dedicated GPU hardware which serves your models for training and inference. You don't need to start with C++, CUDA, and Maths, but: > Components of a GPU and roles > How a token flows through hardware > Compute vs memory, the two limits > What FLOPS means in practice > Training vs inference hardware needs > What is CUDA Kernels > Calculating GPU memory from model size > Quantization to fit bigger models > Multi-GPU and splitting large models Hugging Face has the feature of calculating FLOps and recommending GPUs for your use case. You must explore. In terms of inference engineering, GPU understanding are much helpful your platform performance and handling multiple requests. Read this article for complete breakdown of GPUs for AI Engineers doesn't require any pre-requisite.

Understanding GPUs will change your AI Engineering journey beyond agents and LLM. Rise of LLMs and deep learning introduced dedicated GPU hardware which serves your models for training and inference. You don't need to start with C++, CUDA, and Maths, but: > Components of a GPU and roles > How a token flows through hardware > Compute vs memory, the two limits > What FLOPS means in practice > Training vs inference hardware needs > What is CUDA Kernels > Calculating GPU memory from model size > Quantization to fit bigger models > Multi-GPU and splitting large models Hugging Face has the feature of calculating FLOps and recommending GPUs for your use case. You must explore. In terms of inference engineering, GPU understanding are much helpful your platform performance and handling multiple requests. Read this article for complete breakdown of GPUs for AI Engineers doesn't require any pre-requisite.

Karan🧋

33,453 Aufrufe • vor 1 Monat

I wrote GPU kernels for GPT2 124M inference engine from scratch in ~1500 lines of CUDA. Here I explain it simply.

I wrote GPU kernels for GPT2 124M inference engine from scratch in ~1500 lines of CUDA. Here I explain it simply.

ninjaHattori1

160,092 Aufrufe • vor 2 Monaten

Introducing The AI CUDA Engineer: An agentic AI system that automates the production of highly optimized CUDA kernels. The AI CUDA Engineer can produce highly optimized CUDA kernels, reaching 10-100x speedup over common machine learning operations in PyTorch. Our system is also able to produce highly optimized CUDA kernels that are much faster than existing CUDA kernels commonly used in production. We believe that fundamentally, AI systems can and should be as resource-efficient as the human brain, and that the best path to achieve this efficiency is to use AI to make AI more efficient! We are excited to publish our paper, The AI CUDA Engineer: Agentic CUDA Kernel Discovery, Optimization and Composition. We also release a dataset of over 17,000 verified CUDA kernels produced by The AI CUDA Engineer. Paper: Kernel Archive Webpage: HuggingFace Dataset: The AI CUDA Engineer utilizes evolutionary LLM-driven code optimization to autonomously improve the runtime of machine learning operations. Our system is not only able to convert PyTorch code into CUDA kernels, but through the use of evolution, it can also optimize the runtime performance of CUDA kernels, fuse multiple operations, and even discover novel solutions for writing efficient CUDA operations by learning from past innovations! We believe The AI CUDA Engineer opens a new era of AI-driven acceleration of AI and automated inference time optimization. We (Robert Lange, Aaditya Prasad 🇺🇸, sssss, Maxence Faldor, Yujin Tang, hardmaru) are excited to continue Sakana AI's mission of leveraging AI to improve AI.

Introducing The AI CUDA Engineer: An agentic AI system that automates the production of highly optimized CUDA kernels. The AI CUDA Engineer can produce highly optimized CUDA kernels, reaching 10-100x speedup over common machine learning operations in PyTorch. Our system is also able to produce highly optimized CUDA kernels that are much faster than existing CUDA kernels commonly used in production. We believe that fundamentally, AI systems can and should be as resource-efficient as the human brain, and that the best path to achieve this efficiency is to use AI to make AI more efficient! We are excited to publish our paper, The AI CUDA Engineer: Agentic CUDA Kernel Discovery, Optimization and Composition. We also release a dataset of over 17,000 verified CUDA kernels produced by The AI CUDA Engineer. Paper: Kernel Archive Webpage: HuggingFace Dataset: The AI CUDA Engineer utilizes evolutionary LLM-driven code optimization to autonomously improve the runtime of machine learning operations. Our system is not only able to convert PyTorch code into CUDA kernels, but through the use of evolution, it can also optimize the runtime performance of CUDA kernels, fuse multiple operations, and even discover novel solutions for writing efficient CUDA operations by learning from past innovations! We believe The AI CUDA Engineer opens a new era of AI-driven acceleration of AI and automated inference time optimization. We (Robert Lange, Aaditya Prasad 🇺🇸, sssss, Maxence Faldor, Yujin Tang, hardmaru) are excited to continue Sakana AI's mission of leveraging AI to improve AI.

Sakana AI

1,161,292 Aufrufe • vor 1 Jahr

Ex-vLLM core contributor explained how to make LLM inference 10x cheaper in 34 minutes - better than $3000 inference optimization bootcamps. request comes in -> check LMCache -> hit? load KV cache from CPU/SSD/remote -> skip prefill -> serve. That loop is why Bloomberg and other production stacks now push 300 terabytes of KV cache per week. LMCache + vLLM + CPU/SSD/remote storage + zero-copy CUDA kernels - that's the stack. Watch and save it, then wire the KV-offload into your inference stack.

Ex-vLLM core contributor explained how to make LLM inference 10x cheaper in 34 minutes - better than $3000 inference optimization bootcamps. request comes in -> check LMCache -> hit? load KV cache from CPU/SSD/remote -> skip prefill -> serve. That loop is why Bloomberg and other production stacks now push 300 terabytes of KV cache per week. LMCache + vLLM + CPU/SSD/remote storage + zero-copy CUDA kernels - that's the stack. Watch and save it, then wire the KV-offload into your inference stack.

h100envy

61,470 Aufrufe • vor 4 Monaten

AI that responds in under 100ms doesn't happen by accident. Together AI VP of Kernels Dan Fu shares how they use NVIDIA CUDA, TensorRT-LLM, Dynamo and Together ATLAS on NVIDIA Blackwell to power ultra-low-latency inference and long context code generation for Cursor. From building megakernels that fuse an entire model's forward pass into a single launch to enabling real-time voice agents, Together AI is optimizing latency at every layer of the stack.

AI that responds in under 100ms doesn't happen by accident. Together AI VP of Kernels Dan Fu shares how they use NVIDIA CUDA, TensorRT-LLM, Dynamo and Together ATLAS on NVIDIA Blackwell to power ultra-low-latency inference and long context code generation for Cursor. From building megakernels that fuse an entire model's forward pass into a single launch to enabling real-time voice agents, Together AI is optimizing latency at every layer of the stack.

NVIDIA AI Infrastructure

20,770 Aufrufe • vor 4 Monaten

Luminal ( is creating PyTorch for Production – an ML compiler that generates blazingly fast CUDA kernels and makes deploying to production one line of code. Congrats on the launch, Jake Stevens, Joe Fioti, and Matthew Gunton!

Luminal ( is creating PyTorch for Production – an ML compiler that generates blazingly fast CUDA kernels and makes deploying to production one line of code. Congrats on the launch, Jake Stevens, Joe Fioti, and Matthew Gunton!

Y Combinator

98,496 Aufrufe • vor 1 Jahr

PyTorch 2.13 brings FlexAttention to Apple Silicon, cuts peak memory by up to 4× for large-vocabulary models with nn.LinearCrossEntropyLoss, and updates distributed training, compilation, profiling, and on-device inference. Our live Q&A examined CUDA version support, CuTeDSL in TorchInductor, Python 3.15, ExecuTorch, torch.compile, and plans for PyTorch 2.14. Andrey Talman (Meta), albanD (Meta), and Piotr Bialecki (NVIDIA) joined moderator Chris Gottbrath (Gottbrath Technologies). This clip explains how FlexAttention compiles kernels specialized for the requested masking pattern, allowing them to skip unnecessary computation. It also covers speedups of up to approximately 12× over SDPA in tested sparse configurations and a deterministic backward path on CUDA that removes one source of nondeterminism in gradient computation. 🔗 Watch the full PyTorch 2.13 Release Live Q&A:

PyTorch 2.13 brings FlexAttention to Apple Silicon, cuts peak memory by up to 4× for large-vocabulary models with nn.LinearCrossEntropyLoss, and updates distributed training, compilation, profiling, and on-device inference. Our live Q&A examined CUDA version support, CuTeDSL in TorchInductor, Python 3.15, ExecuTorch, torch.compile, and plans for PyTorch 2.14. Andrey Talman (Meta), albanD (Meta), and Piotr Bialecki (NVIDIA) joined moderator Chris Gottbrath (Gottbrath Technologies). This clip explains how FlexAttention compiles kernels specialized for the requested masking pattern, allowing them to skip unnecessary computation. It also covers speedups of up to approximately 12× over SDPA in tested sparse configurations and a deterministic backward path on CUDA that removes one source of nondeterminism in gradient computation. 🔗 Watch the full PyTorch 2.13 Release Live Q&A:

PyTorch

18,591 Aufrufe • vor 3 Monaten

Already running an inference engine? So where does NVIDIA Dynamo fit in? In five minutes, we break down how Dynamo sits around engines like SGLang, vLLM and TensorRT-LLM to scale inference across GPUs and nodes. Full video in the comments 🔽

Already running an inference engine? So where does NVIDIA Dynamo fit in? In five minutes, we break down how Dynamo sits around engines like SGLang, vLLM and TensorRT-LLM to scale inference across GPUs and nodes. Full video in the comments 🔽

NVIDIA AI

30,908 Aufrufe • vor 2 Monaten

LLM inference speed with vs. without KV caching: (learn how and why it works below)

LLM inference speed with vs. without KV caching: (learn how and why it works below)

Avi Chawla

396,012 Aufrufe • vor 7 Monaten

LLM inference speed with vs. without KV caching: (learn how and why it works below)

LLM inference speed with vs. without KV caching: (learn how and why it works below)

Daily Dose of Data Science

59,218 Aufrufe • vor 6 Monaten

Ex-NVIDIA engineer who built Unsloth explained RL, kernels, reasoning, quantization, and agents in 2 hours 42 minutes - better than $5000 fine-tuning bootcamps. pick the base model -> write triton kernels for 2x faster fine-tune -> quantize to 4-bit -> run GRPO/DPO -> ship a reasoning model on your single GPU. That loop is why Unsloth is the default way to fine-tune Llama, Qwen, Gemma, and Phi on hardware you already own. Unsloth + Triton kernels + 4-bit quantization + GRPO/DPO + single-GPU fine-tuning - that's the stack. Watch and save it, then fine-tune your first model tonight.

Ex-NVIDIA engineer who built Unsloth explained RL, kernels, reasoning, quantization, and agents in 2 hours 42 minutes - better than $5000 fine-tuning bootcamps. pick the base model -> write triton kernels for 2x faster fine-tune -> quantize to 4-bit -> run GRPO/DPO -> ship a reasoning model on your single GPU. That loop is why Unsloth is the default way to fine-tune Llama, Qwen, Gemma, and Phi on hardware you already own. Unsloth + Triton kernels + 4-bit quantization + GRPO/DPO + single-GPU fine-tuning - that's the stack. Watch and save it, then fine-tune your first model tonight.

h100envy

498,100 Aufrufe • vor 4 Monaten

At #PyTorchCon North America 2026, Liz Li and Shekhar of AMD will discuss their work enabling scalable MXFP8 pretraining on MI355X GPUs. Liz and Shekhar will cover optimization of key MXFP8 operators in TorchAO and how those kernels were integrated with TorchTitan to build an end-to-end upstream PyTorch training stack for large language models. Their session, “Scaling MXFP8 Pretraining on 1K+ AMD Instinct MI355X: TorchAO Kernels and TorchTitan Training,” will also cover Triton and FlyDSL implementations and lessons from numerical validation, convergence, and performance tuning. Join us in San Jose on October 20-21. Register today:

At #PyTorchCon North America 2026, Liz Li and Shekhar of AMD will discuss their work enabling scalable MXFP8 pretraining on MI355X GPUs. Liz and Shekhar will cover optimization of key MXFP8 operators in TorchAO and how those kernels were integrated with TorchTitan to build an end-to-end upstream PyTorch training stack for large language models. Their session, “Scaling MXFP8 Pretraining on 1K+ AMD Instinct MI355X: TorchAO Kernels and TorchTitan Training,” will also cover Triton and FlyDSL implementations and lessons from numerical validation, convergence, and performance tuning. Join us in San Jose on October 20-21. Register today:

PyTorch

16,137 Aufrufe • vor 1 Monat

SGLang-Diffusion is a high-performance serving framework for diffusion models, designed for both large-scale offline generation and latency-sensitive real-time inference. At PyTorch Conference North America 2026, Yihao Wang and Kevin Mi of RadixArk will present the design of SGLang-Diffusion and the PyTorch-based inference optimizations behind it, including efficient request scheduling, memory management, batching strategies, and execution-path optimizations for diffusion pipelines. Join us in San Jose, October 20-21: #PyTorchCon

SGLang-Diffusion is a high-performance serving framework for diffusion models, designed for both large-scale offline generation and latency-sensitive real-time inference. At PyTorch Conference North America 2026, Yihao Wang and Kevin Mi of RadixArk will present the design of SGLang-Diffusion and the PyTorch-based inference optimizations behind it, including efficient request scheduling, memory management, batching strategies, and execution-path optimizations for diffusion pipelines. Join us in San Jose, October 20-21: #PyTorchCon

PyTorch

17,471 Aufrufe • vor 1 Monat

Interested in optimizing Mixture of Experts LLM inference on ARM CPUs? In his talk at PyTorch Conference North America, Maajid Khan from Fujitsu Research India will explore efficient MoE LLM inference using vLLM and OpenVINO, sharing practical strategies for running these models effectively on ARM architectures. Register and join us in San Jose, October 20-21: #PyTorchCon

Interested in optimizing Mixture of Experts LLM inference on ARM CPUs? In his talk at PyTorch Conference North America, Maajid Khan from Fujitsu Research India will explore efficient MoE LLM inference using vLLM and OpenVINO, sharing practical strategies for running these models effectively on ARM architectures. Register and join us in San Jose, October 20-21: #PyTorchCon

PyTorch

16,448 Aufrufe • vor 1 Monat

>build a 1.5B VLA on top of MiniCPM-V 4.6 >compete with π0.5 and Qwen-VLA >use roughly half the visual-processing compute >remember previous steps without blowing up inference cost >pull the network plug >let the robot keep tracking — fully local, fully onboard >with CUDA Graph optimization and custom Triton fused kernels, MiniCPM-Robot inference throughput goes from 10 Hz → 33 Hz, and up to 36 Hz on the same NVIDIA H20 >release the models and framework as open source so why does robotics still rely on bigger models and the cloud? OpenBMB just dropped MiniCPM-Robot at WAIC two models + one inference framework 🧵

>build a 1.5B VLA on top of MiniCPM-V 4.6 >compete with π0.5 and Qwen-VLA >use roughly half the visual-processing compute >remember previous steps without blowing up inference cost >pull the network plug >let the robot keep tracking — fully local, fully onboard >with CUDA Graph optimization and custom Triton fused kernels, MiniCPM-Robot inference throughput goes from 10 Hz → 33 Hz, and up to 36 Hz on the same NVIDIA H20 >release the models and framework as open source so why does robotics still rely on bigger models and the cloud? OpenBMB just dropped MiniCPM-Robot at WAIC two models + one inference framework 🧵

arc.

48,279 Aufrufe • vor 3 Monaten

I sat down with Silas Alberti, Head of Research Cognition, to chat about the intersection of RL and inference -- from training frontier coding models to running inference at scale. Watch till the end for bonus doggo! :o 0:00 — Intros 1:04 — What's hardest to get right in an RL run 3:39 — How Cognition started using Modal 4:03 — What to weigh when buying inference 4:58 — The latency/throughput Pareto frontier 7:14 — DFlash reveal (!!) 8:44 — Why faster runtime matters 9:56 — Memory-bound vs. compute-bound 10:57 — Tree-based speculation 14:20 — Hot take: RL and inference optimization converging 15:28 — Online training of the speculator 16:22 — "Auto inference" 17:27 — Agents competing on optimization problems 18:27 — What inference providers still get wrong

I sat down with Silas Alberti, Head of Research Cognition, to chat about the intersection of RL and inference -- from training frontier coding models to running inference at scale. Watch till the end for bonus doggo! :o 0:00 — Intros 1:04 — What's hardest to get right in an RL run 3:39 — How Cognition started using Modal 4:03 — What to weigh when buying inference 4:58 — The latency/throughput Pareto frontier 7:14 — DFlash reveal (!!) 8:44 — Why faster runtime matters 9:56 — Memory-bound vs. compute-bound 10:57 — Tree-based speculation 14:20 — Hot take: RL and inference optimization converging 15:28 — Online training of the speculator 16:22 — "Auto inference" 17:27 — Agents competing on optimization problems 18:27 — What inference providers still get wrong

gongy

82,724 Aufrufe • vor 3 Monaten

At PyTorch Conference North America 2026, hear directly from and connect with the people working on PyTorch, vLLM, DeepSpeed, ray, Helion, and Safetensors, alongside experts working across the AI stack. Keynote speaker Simon Mo (vLLM, Inferact) says vLLM’s goal is to become “the easiest to use and most efficient inference engine.” At #PyTorchCon NA 2025, he spoke about the difficulty and cost of using large language models and continuing to push inference efficiency forward. PyTorch conferences are the open source AI community’s town square, where what’s next gets decided. Register by September 4 to save on your conference pass:

At PyTorch Conference North America 2026, hear directly from and connect with the people working on PyTorch, vLLM, DeepSpeed, ray, Helion, and Safetensors, alongside experts working across the AI stack. Keynote speaker Simon Mo (vLLM, Inferact) says vLLM’s goal is to become “the easiest to use and most efficient inference engine.” At #PyTorchCon NA 2025, he spoke about the difficulty and cost of using large language models and continuing to push inference efficiency forward. PyTorch conferences are the open source AI community’s town square, where what’s next gets decided. Register by September 4 to save on your conference pass:

PyTorch

22,334 Aufrufe • vor 2 Monaten

ERRC: Entropy-Reinvested Residual Correction for Tensor-Parallel LLM Communication PyTorch Foundation Ambassador Abdulsalam Bande will present a poster on ERRC (Entropy-Reinvested Residual Correction for Tensor-Parallel LLM Communication) at PyTorch Conference North America 2026. ERRC focuses on optimizing GPU data exchange during LLM inference. By compressing inter-GPU communication and using the saved bandwidth to mitigate quantization errors, ERRC makes multi-GPU inference significantly more efficient. Register and join us to San Jose, October 20-21: #PyTorchCon Abdulsalam Bande

ERRC: Entropy-Reinvested Residual Correction for Tensor-Parallel LLM Communication PyTorch Foundation Ambassador Abdulsalam Bande will present a poster on ERRC (Entropy-Reinvested Residual Correction for Tensor-Parallel LLM Communication) at PyTorch Conference North America 2026. ERRC focuses on optimizing GPU data exchange during LLM inference. By compressing inter-GPU communication and using the saved bandwidth to mitigate quantization errors, ERRC makes multi-GPU inference significantly more efficient. Register and join us to San Jose, October 20-21: #PyTorchCon Abdulsalam Bande

PyTorch

16,886 Aufrufe • vor 1 Monat

PyTorch core engineer at Meta turned CUDA kernel writing into a sport in 13 minutes - better than $1500 GPU programming bootcamps. profile the kernel -> find the bottleneck -> rewrite -> benchmark -> merge the winning code into PyTorch. That loop is how the open community now beats hand-tuned vendor kernels. GPU MODE community + KernelBot competition + winning kernel merged into the framework - that's the stack. Watch it, then steal the loop below.

h100envy

35,390 Aufrufe • vor 4 Monaten