正在加载视频...

视频加载失败

加载此视频时出现问题。这可能是由于临时网络问题，或视频可能不可用。

Training LFMs at scale means solving parallelism across every layer of the architecture. And not all layers are the same. Our CTO Mathias Lechner (Mathias Lechner) sits down with Liquid's founding engineer Paul Pak (Paul Pak) to talk training infrastructure: Data, tensor, pipeline, expert, and context parallelism, and how... show more

Liquid AI

29,833 subscribers

17,125 次观看 • 25 天前 •via X (Twitter)

Anya Rossi• Live Now

Private livecam show

0 条评论

暂无评论

原始帖子的评论将显示在这里

相关视频

Building a model is just the start. Post-training makes it useful. Our CTO Mathias Lechner (Mathias Lechner) sits down for a conversation with Maxime Labonne (Maxime Labonne), our head of post-training, on the pipeline that takes a base model from autocomplete to something that can reason and follow instructions.

Building a model is just the start. Post-training makes it useful. Our CTO Mathias Lechner (Mathias Lechner) sits down for a conversation with Maxime Labonne (Maxime Labonne), our head of post-training, on the pipeline that takes a base model from autocomplete to something that can reason and follow instructions.

Liquid AI

25,877 次观看 • 1 个月前

The bottleneck in LLM inference isn't compute. It's how fast you can move the weights. Our CTO Mathias Lechner, Mathias Lechner, joins Piotr Mazurek, Piotr Mazurek (in SF 🌉), from our inference team, to discuss what actually limits token throughput and how we're optimizing for it.

The bottleneck in LLM inference isn't compute. It's how fast you can move the weights. Our CTO Mathias Lechner, Mathias Lechner, joins Piotr Mazurek, Piotr Mazurek (in SF 🌉), from our inference team, to discuss what actually limits token throughput and how we're optimizing for it.

Liquid AI

21,393 次观看 • 1 个月前

Geometric Context Transformer for Streaming 3D Reconstruction Contributions: • We introduce LingBot-Map, a streaming 3D foundation model built around Geometric Context Attention (GCA), which maintains three complementary context types – anchor, pose-reference window, and trajectory memory – for efficient and consistent long-sequence streaming inference. • We propose an efficient training recipe based on progressive training and context parallelism with a relative loss formulation for stable long-sequence optimization. • We demonstrate that LingBot-Map achieves state-of-the-art performance on multiple benchmarks (Oxford Spires, Tanks and Temples, ETH3D, and 7-Scenes), significantly outperforming existing streaming approaches in reconstruction quality and inference speed.

Geometric Context Transformer for Streaming 3D Reconstruction Contributions: • We introduce LingBot-Map, a streaming 3D foundation model built around Geometric Context Attention (GCA), which maintains three complementary context types – anchor, pose-reference window, and trajectory memory – for efficient and consistent long-sequence streaming inference. • We propose an efficient training recipe based on progressive training and context parallelism with a relative loss formulation for stable long-sequence optimization. • We demonstrate that LingBot-Map achieves state-of-the-art performance on multiple benchmarks (Oxford Spires, Tanks and Temples, ETH3D, and 7-Scenes), significantly outperforming existing streaming approaches in reconstruction quality and inference speed.

MrNeRF

24,623 次观看 • 2 个月前

we're turning University of Waterloo into a supercomputer! arceus is a cross-device distributed compute network for training large models, using model/tensor/pipeline parallelism. you can train anything, from deep neural networks to language models on the network. oss & deployment soon!

we're turning University of Waterloo into a supercomputer! arceus is a cross-device distributed compute network for training large models, using model/tensor/pipeline parallelism. you can train anything, from deep neural networks to language models on the network. oss & deployment soon!

rajan agarwal

35,534 次观看 • 1 年前

New course: Efficient Inference with SGLang: Text and Image Generation, built in partnership with LMSys LMSYS Org and RadixArk RadixArk, and taught by Richard Chen Richard Chen, a Member of Technical Staff at RadixArk. Running LLMs in production is expensive, and much of that cost comes from redundant computation. This short course teaches you to eliminate that waste using SGLang, an open-source inference framework that caches computation already done and reuses it across future requests. When ten users share the same system prompt, SGLang processes it once, not ten times. The speedups compound quickly, especially when there's a lot of shared context across requests. Skills you'll gain: - Implement a KV cache from scratch to eliminate redundant computation within a single request - Scale caching across users and requests with RadixAttention, so shared context is only processed once - Accelerate image generation with diffusion models using SGLang's caching and multi-GPU parallelism Join and learn to make LLM inference faster and more cost-efficient at scale!

New course: Efficient Inference with SGLang: Text and Image Generation, built in partnership with LMSys LMSYS Org and RadixArk RadixArk, and taught by Richard Chen Richard Chen, a Member of Technical Staff at RadixArk. Running LLMs in production is expensive, and much of that cost comes from redundant computation. This short course teaches you to eliminate that waste using SGLang, an open-source inference framework that caches computation already done and reuses it across future requests. When ten users share the same system prompt, SGLang processes it once, not ten times. The speedups compound quickly, especially when there's a lot of shared context across requests. Skills you'll gain: - Implement a KV cache from scratch to eliminate redundant computation within a single request - Scale caching across users and requests with RadixAttention, so shared context is only processed once - Accelerate image generation with diffusion models using SGLang's caching and multi-GPU parallelism Join and learn to make LLM inference faster and more cost-efficient at scale!

Andrew Ng

98,292 次观看 • 2 个月前

GLM5 running on a cluster of 4 Mac Studios using tensor parallelism thanks to MLX distributed. Yes, those are the Alex drawers.

GLM5 running on a cluster of 4 Mac Studios using tensor parallelism thanks to MLX distributed. Yes, those are the Alex drawers.

Alex Ziskind

20,594 次观看 • 4 个月前

•one for the parallelism in #meemsemohabbat [ishq hai- ft. talha and roshi]

•one for the parallelism in #meemsemohabbat [ishq hai- ft. talha and roshi]

sarah

21,366 次观看 • 1 年前

the parallelism that there is between these two in live, #taekook there is so much subtlety and there are still pple who doubt how close they are?

the parallelism that there is between these two in live, #taekook there is so much subtlety and there are still pple who doubt how close they are?

⭒

40,627 次观看 • 3 年前

PWR Chain brings internet-like scalability to blockchain. 🌐 Here’s how 👇 • Others: Parallelism at smart contract level within one VM • PWR: Parallelism at VM and software application level - Supports any VM or applications - Similar to parallel servers on the internet - Efficient inter-application communication

PWR Chain brings internet-like scalability to blockchain. 🌐 Here’s how 👇 • Others: Parallelism at smart contract level within one VM • PWR: Parallelism at VM and software application level - Supports any VM or applications - Similar to parallel servers on the internet - Efficient inter-application communication

PWR Chain

94,409 次观看 • 1 年前

🚀 Excited to release LongLive 2.0! 🎬 An end-to-end infrastructure for long video generation, with FP4 and parallelism at the core of both training and inference. ⚡45.7 FPS generation speed on 5B model⚡ ✨ LongLive 2.0 supports real-video training, few-step distillation, multi-shot training/inference, sequence-parallel acceleration, NVFP4 KV cache, and async VAE decoding deployment. 🧩 To our knowledge, this is the first open-source 4-bit long video generation infra that covers both training and inference. 🙌 Welcome to check it out, try it, and share feedback! 🔗 Code: 📰 Paper: 🎥 Demo: #LongVideoGeneration #VideoGeneration #Realtime #AIInfra #EfficientAI #FP4 #Parallel #NVIDIA

🚀 Excited to release LongLive 2.0! 🎬 An end-to-end infrastructure for long video generation, with FP4 and parallelism at the core of both training and inference. ⚡45.7 FPS generation speed on 5B model⚡ ✨ LongLive 2.0 supports real-video training, few-step distillation, multi-shot training/inference, sequence-parallel acceleration, NVFP4 KV cache, and async VAE decoding deployment. 🧩 To our knowledge, this is the first open-source 4-bit long video generation infra that covers both training and inference. 🙌 Welcome to check it out, try it, and share feedback! 🔗 Code: 📰 Paper: 🎥 Demo: #LongVideoGeneration #VideoGeneration #Realtime #AIInfra #EfficientAI #FP4 #Parallel #NVIDIA

Yukang Chen

58,301 次观看 • 1 个月前

In my experience, robot 'generalists' are often jacks of all trades but masters of none. In training across multiple tasks and environments, robot policies fail to generalize robustly and effectively to each particular test setting. What if at test time, we non-parametrically *retrieved* “relevant” data from the training set and used it to significantly improve the performance of few-shot imitation learning to be robust to various test time scenes. Notably, we are *not* collecting lots of new data, just training more on sub-components of the same training data! Now, we’re certainly not the first to suggest retrieval, but in our new work - STRAP, we show how retrieving relevant *sub-trajectories* from offline datasets can significantly increase data reuse across tasks, when paired with an appropriate metric space. A 🧵 (1/7)

In my experience, robot 'generalists' are often jacks of all trades but masters of none. In training across multiple tasks and environments, robot policies fail to generalize robustly and effectively to each particular test setting. What if at test time, we non-parametrically retrieved “relevant” data from the training set and used it to significantly improve the performance of few-shot imitation learning to be robust to various test time scenes. Notably, we are not collecting lots of new data, just training more on sub-components of the same training data! Now, we’re certainly not the first to suggest retrieval, but in our new work - STRAP, we show how retrieving relevant sub-trajectories from offline datasets can significantly increase data reuse across tasks, when paired with an appropriate metric space. A 🧵 (1/7)

Abhishek Gupta

12,045 次观看 • 1 年前

GLM-4.7-8bit (350GB) running at 19 toks/s on two M3 Ultra 512GB using Tensor Parallelism with EXO - MLX, versus 14 toks/s with single node. 🚀 Now context benchmarking & then OpenCode tests 🔥 Note: this is from sources, I had to change things to run it.

GLM-4.7-8bit (350GB) running at 19 toks/s on two M3 Ultra 512GB using Tensor Parallelism with EXO - MLX, versus 14 toks/s with single node. 🚀 Now context benchmarking & then OpenCode tests 🔥 Note: this is from sources, I had to change things to run it.

Ivan Fioravanti ᯅ

327,687 次观看 • 6 个月前

Get to know River, our internal agent that among other things drastically reduces the cost of context gathering across the company. No more digging through docs, Slack, and requiring people to spread context. A new kind of information architecture. Talked about this with tobi lutke on our Context episode:

Get to know River, our internal agent that among other things drastically reduces the cost of context gathering across the company. No more digging through docs, Slack, and requiring people to spread context. A new kind of information architecture. Talked about this with tobi lutke on our Context episode:

Jessica Hertz

13,909 次观看 • 1 个月前

𝐓𝐡𝐞 𝐒𝐀𝐏 𝐀𝐫𝐜𝐡𝐢𝐭𝐞𝐜𝐭𝐮𝐫𝐞 𝐂𝐞𝐧𝐭𝐞𝐫 𝐢𝐬 𝐧𝐨𝐰 𝐥𝐢𝐯𝐞 𝐨𝐧 𝐭𝐡𝐞 𝐒𝐀𝐏 𝐋𝐞𝐚𝐫𝐧𝐢𝐧𝐠 𝐂𝐞𝐧𝐭𝐞𝐫: More details in the post Built by our team at the Office of the CTO, SAP, the Architecture Center is designed to scale how we build, publish, and govern reference solution architectures across the SAP ecosystem, with open collaboration at its core. Anirban Navya Khurana SAP LABS INDIA Sindhu Gangadharan

𝐓𝐡𝐞 𝐒𝐀𝐏 𝐀𝐫𝐜𝐡𝐢𝐭𝐞𝐜𝐭𝐮𝐫𝐞 𝐂𝐞𝐧𝐭𝐞𝐫 𝐢𝐬 𝐧𝐨𝐰 𝐥𝐢𝐯𝐞 𝐨𝐧 𝐭𝐡𝐞 𝐒𝐀𝐏 𝐋𝐞𝐚𝐫𝐧𝐢𝐧𝐠 𝐂𝐞𝐧𝐭𝐞𝐫: More details in the post Built by our team at the Office of the CTO, SAP, the Architecture Center is designed to scale how we build, publish, and govern reference solution architectures across the SAP ecosystem, with open collaboration at its core. Anirban Navya Khurana SAP LABS INDIA Sindhu Gangadharan

PVN PavanKumar

251,687 次观看 • 1 年前

Our approach to AI infra is simple: build the most fungible and flexible fleet to meet the real world's needs across inference and training as Scott Guthrie shared with Alex Kantrowitz. And we are already doing it at scale today, as we power the biggest AI workloads like Copilot and ChatGPT, APIs that power 3P products & enterprise workloads and high scale training.

Our approach to AI infra is simple: build the most fungible and flexible fleet to meet the real world's needs across inference and training as Scott Guthrie shared with Alex Kantrowitz. And we are already doing it at scale today, as we power the biggest AI workloads like Copilot and ChatGPT, APIs that power 3P products & enterprise workloads and high scale training.

Satya Nadella

114,478 次观看 • 9 个月前

Concurrency vs Parallelism

Concurrency vs Parallelism

Alex Xu

40,134 次观看 • 10 个月前

SF Tensor (San Francisco Tensor Company) is building infrastructure for training foundation models, letting researchers focus on research and forget that the infrastructure layer even exists while saving 70% in training costs. Congrats on the launch, Ben Koska, Tom Koska, and Luk Koska!

SF Tensor (San Francisco Tensor Company) is building infrastructure for training foundation models, letting researchers focus on research and forget that the infrastructure layer even exists while saving 70% in training costs. Congrats on the launch, Ben Koska, Tom Koska, and Luk Koska!

Y Combinator

10,721 次观看 • 7 个月前

Tonight in SF – Pathway (www.pathway.com) brings the inventors to settle it in the ring. 🥊 Transformer vs Post-Transformer: The Deciding Round. Featuring: Lukasz Kaiser, Adrian Kosowski, Mathias Lechner, Llion Jones, dex and myself!

Tonight in SF – Pathway (www.pathway.com) brings the inventors to settle it in the ring. 🥊 Transformer vs Post-Transformer: The Deciding Round. Featuring: Lukasz Kaiser, Adrian Kosowski, Mathias Lechner, Llion Jones, dex and myself!

Zuzanna Stamirowska

153,276 次观看 • 1 个月前

Enterprise AI hits its ceiling at the data layer. Storage bandwidth, data orchestration, and GPU efficiency are where the token economy is actually won or lost. Patrick Moorhead and Daniel Newman sat down with Dell Technologies ISG CTO Ihab Tarazi and Jason Hardy VP NVIDIA at Dell Technologies World 2026, to break down the Dell AI Data Platform, Exascale Storage, and what co-engineering actually means for enterprise AI at scale.

Enterprise AI hits its ceiling at the data layer. Storage bandwidth, data orchestration, and GPU efficiency are where the token economy is actually won or lost. Patrick Moorhead and Daniel Newman sat down with Dell Technologies ISG CTO Ihab Tarazi and Jason Hardy VP NVIDIA at Dell Technologies World 2026, to break down the Dell AI Data Platform, Exascale Storage, and what co-engineering actually means for enterprise AI at scale.

Six Five Media

91,546 次观看 • 1 个月前