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。

Introducing Halo, the best framework for post-training of open-source models. Halo delivers up to 2.8x the throughput of stock TRL with less peak memory, while models stay in their native HuggingFace format. Star us on GitHub:

White Circle

2,086 subscribers

2,252,490 次观看 • 6 天前 •via X (Twitter)

新闻政治教育科学技术

Anya Rossi• Live Now

Private livecam show

56 条评论

White Circle 的头像

White Circle6 天前

Every model we train at White Circle now uses Halo. The same codebase runs LoRA on a 24 GB GPU, multi-node training on B300s, and async RL. Technical blog:

White Circle 的头像

White Circle6 天前

Other training frameworks often require a separate implementation for each model family. In Halo, adding a new model family takes just 100 lines of connecting a wrapper instead of rewriting the model.

White Circle 的头像

White Circle6 天前

One YAML file contains the model, training method, GPU setup and checkpoint settings. One command starts training.

White Circle 的头像

White Circle6 天前

Halo supports data, tensor, context, expert and expert-tensor parallelism. It also includes optimized attention, grouped-GEMM and fused loss kernels.

White Circle 的头像

White Circle6 天前

Halo is not limited to supervised fine-tuning. It supports async RL and training with external environments. We are excited to partner with the @sgl_project team to make it the primary engine for rollouts.

White Circle 的头像

White Circle6 天前

On @OpenAI gpt-oss-20b, Halo delivered 2.3–2.8x the throughput of stock TRL with less peak memory. Both runs used the same FlashAttention, Liger, fused cross-entropy and grouped-GEMM optimizations.

White Circle 的头像

White Circle6 天前

We partnered with @liquidai to add native Halo support for LFM2.5-8B-A1B. On one B300, Halo was up to 20% faster than next best open-source framework.

White Circle 的头像

White Circle6 天前

We also used Halo to fine-tune @Zai_org GLM-4.7-Flash on 177M tokens of agentic traces. The resulting model improved SWE-rebench-V2 by @nebiusai from 33% to 42%. Halo reached up to 1.63× TRL throughput on the same model precision and data. Model and write-up:

White Circle 的头像

White Circle6 天前

Thanks for reading to the end! Star us on GitHub: Read more:

Daniel Smidstrup 的头像

Daniel Smidstrup6 天前

Open source is the way, and it's only going to get more and more relevant in the coming time! :)

White Circle 的头像

White Circle6 天前

True

leo 的头像

leo6 天前

insane way to kick off the week -- congrats @whitecircle team

White Circle 的头像

White Circle6 天前

Lets go!

neural nets. 的头像

neural nets.6 天前

crazy reading through the docs

Vishal Singh 🥑 的头像

Vishal Singh 🥑6 天前

love that this is open source, congrats team ❤️

Catalin 的头像

Catalin6 天前

Congrats on the awesome work, team! Good to see training infra being open source 👏 Starred the repo to contribute to the growth.

White Circle 的头像

White Circle6 天前

thanks!

Paul Mit 的头像

Paul Mit6 天前

great idea you've got my star, guys

giulia 的头像

giulia6 天前

took me one read to understand exactly who this is for. that almost never happens with infra launches. congrats team!!! starred 👀

Mac 的头像

Mac6 天前

Peak improvement, I will try it from first hand

sui 的头像

sui6 天前

so we get Grok 4.7 along with this today? nice.

White Circle 的头像

White Circle6 天前

Crazy day!

Maxime Labonne 的头像

Maxime Labonne6 天前

Congrats guys!

Virgile RIETSCH 的头像

Virgile RIETSCH6 天前

This is huge !!

Levan Kvirkvelia 的头像

Levan Kvirkvelia6 天前

haloshi framework

𝗕𝗿𝗶𝗮𝗻 𝗥𝗲𝘆 的头像

𝗕𝗿𝗶𝗮𝗻 𝗥𝗲𝘆6 天前

wait what this is incredible

White Circle 的头像

White Circle6 天前

ty Brian! show the repo some love ⭐

ivy 的头像

ivy6 天前

so excited!!!

White Circle 的头像

White Circle6 天前

🥰

Ted | unfair.so 的头像

Ted | unfair.so6 天前

Congrats team!

Soraia 的头像

Soraia6 天前

Great work! You got my star ⭐️

Nidhi Singh 的头像

Nidhi Singh6 天前

love that this is open source, congrats!

Rushil Chopra 的头像

Rushil Chopra6 天前

I have always wanted to train a custom math model, let’s see how this goes!

White Circle 的头像

White Circle6 天前

Keep us posted, any feedback is welcome

AshutoshShrivastava 的头像

AshutoshShrivastava6 天前

love seeing more training infrastructure get open-sourced. congrats on the release!

Ion 的头像

Ion6 天前

Congrats! Impressive what you delivered there 👀

Nikita Andersson 的头像

Nikita Andersson6 天前

@sarapenrique This is fabulous

Hauke 🌱🦀 的头像

Hauke 🌱🦀6 天前

Uh training my own models for my products would be pretty neat. I need better hardware 👀

White Circle 的头像

White Circle6 天前

💚

Prasenjit 的头像

Prasenjit6 天前

oh this is sick actually, well cooked

Kalash 的头像

Kalash6 天前

banger stuff 💥

Dan Kulkov 的头像

Dan Kulkov6 天前

insane start of the week congrats team!

Aditi 的头像

Aditi6 天前

wait… 2.8x faster than stock trl ??

Vasko 的头像

Vasko6 天前

congrats to the team!

Lucas Valbuena 的头像

Lucas Valbuena6 天前

congrats!

White Circle 的头像

White Circle6 天前

Thank you

Sameer 的头像

Sameer6 天前

congrats! def keeping an eye on this one.

Suraj Sharma 的头像

Suraj Sharma6 天前

The interesting part is keeping models in native HuggingFace format while getting that throughput boost.

White Circle 的头像

White Circle6 天前

Worked hard on this

Suraj Sharma 的头像

Suraj Sharma6 天前

LFG 🫡🙌

Jackson 的头像

Jackson6 天前

Really cool. Really unique and impressive numbers :•)

Marco Franzon 的头像

Marco Franzon6 天前

The only limit now is the hardware. Everyone can now train or finetune a model without rewriting it for the training phase. Awesome!

White Circle 的头像

White Circle6 天前

thanks Marco!

Marco Franzon 的头像

Marco Franzon6 天前

Great work !

Deni 的头像

Deni6 天前

Congrats on the launch team! Starred.

Ole Lehmann 的头像

Ole Lehmann6 天前

happy launch day guys"

相关视频

“We have a crisis of open source models in the Western world. Outside of China, there are no good open source models. We don't even have any in the US now. The talent, the capital and the focus to be best-in-class at pre-training, mid-training, post-training is an extremely scarce skillset. The answer for a lot of countries may just be to take an open weights model from China, post-train or fine-tune their own version, and have that be what they start from." Everett Randle Why does the West have such poor open source models and is it a national security threat to rely on open-source Chinese models Demis Hassabis Yann LeCun Aidan Gomez Lilian Weng Jan Leike

“We have a crisis of open source models in the Western world. Outside of China, there are no good open source models. We don't even have any in the US now. The talent, the capital and the focus to be best-in-class at pre-training, mid-training, post-training is an extremely scarce skillset. The answer for a lot of countries may just be to take an open weights model from China, post-train or fine-tune their own version, and have that be what they start from." Everett Randle Why does the West have such poor open source models and is it a national security threat to rely on open-source Chinese models Demis Hassabis Yann LeCun Aidan Gomez Lilian Weng Jan Leike

Harry Stebbings

63,738 次观看 • 3 个月前

3 weeks ago we open-sourced HALO this led to talking with dozens of teams running agents at scale we realized the current agent monitoring tools aren't built for the future that we so clearly see ahead of us today we’re releasing native OpenTelemetry-compatible agent tracing on Inference, powered by the same open-source core behind HALO

3 weeks ago we open-sourced HALO this led to talking with dozens of teams running agents at scale we realized the current agent monitoring tools aren't built for the future that we so clearly see ahead of us today we’re releasing native OpenTelemetry-compatible agent tracing on Inference, powered by the same open-source core behind HALO

Sam Hogan 🇺🇸

39,791 次观看 • 4 个月前

Introducing ambientGPT: an open-source and multimodal MacOS foundation model GUI Run GPT-4o and open-source models with full ambient knowledge of your screen. Foundation models have long been confined to the browser. With ambientGPT, your screen context is directly inferred as part of the query, ensuring you never need to explicitly upload context again! Github:

Introducing ambientGPT: an open-source and multimodal MacOS foundation model GUI Run GPT-4o and open-source models with full ambient knowledge of your screen. Foundation models have long been confined to the browser. With ambientGPT, your screen context is directly inferred as part of the query, ensuring you never need to explicitly upload context again! Github:

Siddharth Sharma

193,018 次观看 • 2 年前

Today we're announcing Base Labs, a dedicated research organization focused on advancing open-source AI. We believe in a healthy, open frontier model ecosystem. To enable this, we are working on: - Blue-sky research on continual learning, the science of RL, and how models learn across their full lifecycle, with every experiment and recipe shared openly. - The BaseHub Data Foundry: the highest-quality open RL environments, training data, and real-world benchmarks, built for anyone to train and benchmark on. - Post-post training: taking open-source models and making them better, safer, and more aligned through continual post-training, built on our research, and deployed with our frontier safety stack so organizations can use open-source models with confidence. - Making models cheaper and more performant through our model performance research. This is a mission-driven research effort, not a commercial product. We believe the health of the open-source AI ecosystem matters and that the best way to advance it is to do science in the open. We’re hiring engineers, researchers, and research fellows to advance this mission.

Today we're announcing Base Labs, a dedicated research organization focused on advancing open-source AI. We believe in a healthy, open frontier model ecosystem. To enable this, we are working on: - Blue-sky research on continual learning, the science of RL, and how models learn across their full lifecycle, with every experiment and recipe shared openly. - The BaseHub Data Foundry: the highest-quality open RL environments, training data, and real-world benchmarks, built for anyone to train and benchmark on. - Post-post training: taking open-source models and making them better, safer, and more aligned through continual post-training, built on our research, and deployed with our frontier safety stack so organizations can use open-source models with confidence. - Making models cheaper and more performant through our model performance research. This is a mission-driven research effort, not a commercial product. We believe the health of the open-source AI ecosystem matters and that the best way to advance it is to do science in the open. We’re hiring engineers, researchers, and research fellows to advance this mission.

Base Labs

239,395 次观看 • 24 天前

.NVIDIA hand delivered a pre-production unit of the Dell Technologies Pro Max with GB300 to my house. 100lbs beast with 750GB+ of unified memory to power the best open-source models in the world. What should I test first?

.NVIDIA hand delivered a pre-production unit of the Dell Technologies Pro Max with GB300 to my house. 100lbs beast with 750GB+ of unified memory to power the best open-source models in the world. What should I test first?

Matthew Berman

255,703 次观看 • 6 个月前

✨ Personal AI is the next computing platform. AI is shifting from something you access to something you build with, locally, at the edge, and across systems. We’re unlocking new possibilities for developers: • AMD Ryzen AI Halo, a local-first developer system, preorder starting in June, develop AI without limits on your desk • Gorgon Halo with up to 192GB unified memory, supporting 300B+ parameter models locally, run massive models locally We’re excited to partner with clem 🤗🤗, Co-founder and CEO of Hugging Face, to advance open-source AI for Ryzen AI. Our focus is seamless AI, from model to deployment. Cloud, edge, device. One continuum. Multi-agent systems, local inference at scale, open models as infrastructure. This is the next computing era 🚀

✨ Personal AI is the next computing platform. AI is shifting from something you access to something you build with, locally, at the edge, and across systems. We’re unlocking new possibilities for developers: • AMD Ryzen AI Halo, a local-first developer system, preorder starting in June, develop AI without limits on your desk • Gorgon Halo with up to 192GB unified memory, supporting 300B+ parameter models locally, run massive models locally We’re excited to partner with clem 🤗🤗, Co-founder and CEO of Hugging Face, to advance open-source AI for Ryzen AI. Our focus is seamless AI, from model to deployment. Cloud, edge, device. One continuum. Multi-agent systems, local inference at scale, open models as infrastructure. This is the next computing era 🚀

Jack Huynh

50,015 次观看 • 4 个月前

Most AI video tools stop at prompting, but LTX goes a layer deeper. Introducing the new LTX Trainer. Train LoRAs and IC-LoRAs across video, audio, cross-modal, and reference-conditioned workflows from a single framework. Plus: • New agentic skill • Flexible conditioning • Free IC-LoRAs • Fully open source Most models are generic. The best ones become yours. GitHub: Documentation: HuggingFace:

Most AI video tools stop at prompting, but LTX goes a layer deeper. Introducing the new LTX Trainer. Train LoRAs and IC-LoRAs across video, audio, cross-modal, and reference-conditioned workflows from a single framework. Plus: • New agentic skill • Flexible conditioning • Free IC-LoRAs • Fully open source Most models are generic. The best ones become yours. GitHub: Documentation: HuggingFace:

LTX.io

6,455,177 次观看 • 3 个月前

Next up on First Pass: A conversation with Jesse Zhang from Decagon about how everyone is wrong about open source AI Decagon runs ~90% of their workloads on open source models, however their spend on open models fell to 11% (from 19% a year ago).

Next up on First Pass: A conversation with Jesse Zhang from Decagon about how everyone is wrong about open source AI Decagon runs ~90% of their workloads on open source models, however their spend on open models fell to 11% (from 19% a year ago).

Jamin Ball

15,974 次观看 • 1 个月前

The demand for AI capabilities is outpacing the computational capacity available today. As workloads become more complex, small language models are growing larger, with longer input sequences and rising resource demands. The relentless pursuit of near zero latency is now mission-critical as domain-specific models advance. This challenge inspired us to design something radically different — Strix Halo. In deep co-engineering partnership with HP, we’re incredibly proud to power the new ZBook Ultra G1a, the world’s best mobile workstation, and the Z2 Mini G1a, the world’s best mini workstation. Strix Halo shatters the limits of mobile performance, bringing workstation-level power into thin and light laptops. By unlocking access to 128GB of unified memory between CPU and GPU combined with 256GB/s of memory bandwidth, Strix Halo enables architects, engineers, creatives, and data scientists to tackle massive datasets, complex simulations, and AI workflows without compromise. Strix Halo is the world’s first AI PC processor capable of running 70 billion parameter models. Just five years ago, this kind of inference computing required a room full of servers — today, it fits in your backpack. This is a major step forward in the future of AI PCs. Stay tuned as we continue to break boundaries 🚀 #TogetherWeAdvance AMD

The demand for AI capabilities is outpacing the computational capacity available today. As workloads become more complex, small language models are growing larger, with longer input sequences and rising resource demands. The relentless pursuit of near zero latency is now mission-critical as domain-specific models advance. This challenge inspired us to design something radically different — Strix Halo. In deep co-engineering partnership with HP, we’re incredibly proud to power the new ZBook Ultra G1a, the world’s best mobile workstation, and the Z2 Mini G1a, the world’s best mini workstation. Strix Halo shatters the limits of mobile performance, bringing workstation-level power into thin and light laptops. By unlocking access to 128GB of unified memory between CPU and GPU combined with 256GB/s of memory bandwidth, Strix Halo enables architects, engineers, creatives, and data scientists to tackle massive datasets, complex simulations, and AI workflows without compromise. Strix Halo is the world’s first AI PC processor capable of running 70 billion parameter models. Just five years ago, this kind of inference computing required a room full of servers — today, it fits in your backpack. This is a major step forward in the future of AI PCs. Stay tuned as we continue to break boundaries 🚀 #TogetherWeAdvance AMD

Jack Huynh

11,960 次观看 • 1 年前

introducing laravel moat as an open source maintainer, recent supply chain attacks in the ecosystem made me want a simple cli to audit the security of my GitHub organizations and repositories built in Rust. for any open source project on GitHub

introducing laravel moat as an open source maintainer, recent supply chain attacks in the ecosystem made me want a simple cli to audit the security of my GitHub organizations and repositories built in Rust. for any open source project on GitHub

nunomaduro

81,517 次观看 • 4 个月前

Thank you to Markku Räsänen, CEO of ConfidentialMind for sitting down with us and sharing his thoughts on: - Whether open source models are undervalued - Changes in AI infrastructure & inference - Power consumption & efficiency in Europe - The future of open source AI

Thank you to Markku Räsänen, CEO of ConfidentialMind for sitting down with us and sharing his thoughts on: - Whether open source models are undervalued - Changes in AI infrastructure & inference - Power consumption & efficiency in Europe - The future of open source AI

SambaNova

31,874 次观看 • 1 年前

BREAKING $AMD is ready to ship Ryzen AI Halo 🆕 in June 2026!!!! 🚀🚀🚀 AKA Personal/Local AI system. Ryzen AI Max+ 395 (Strix Halo APU) with 128 GB of unified LPDDR5X memory, 40 RDNA 3.5 GPU compute units, and a full ROCm software stack preloaded for both Windows and Linux. AMD Halo is $2,000-$3,000 vs $NVDA Spark $4,699 The main purpose of the AMD Ryzen AI Halo is to serve as a compact, developer focused workstation for running intensive AI workloads locally, specifically building, testing, prototyping, fine-tuning, and deploying generative AI models (like LLMs) and Agentic AI applications without relying on the cloud. Run models with up to 200 billion parameters on your desk using 128 GB of unified memory (shared across CPU, GPU, and NPU). This eliminates cloud costs, latency, data privacy concerns, and internet dependency. ~This is for AI developers, researchers, engineers, and creators. ~Teams or individuals wanting faster iteration, data privacy, and cost savings compared to cloud APIs. ~Users experimenting with open-source models, fine-tuning, inference, and production-like local deployment Source: AMD CES 2026

BREAKING $AMD is ready to ship Ryzen AI Halo 🆕 in June 2026!!!! 🚀🚀🚀 AKA Personal/Local AI system. Ryzen AI Max+ 395 (Strix Halo APU) with 128 GB of unified LPDDR5X memory, 40 RDNA 3.5 GPU compute units, and a full ROCm software stack preloaded for both Windows and Linux. AMD Halo is $2,000-$3,000 vs $NVDA Spark $4,699 The main purpose of the AMD Ryzen AI Halo is to serve as a compact, developer focused workstation for running intensive AI workloads locally, specifically building, testing, prototyping, fine-tuning, and deploying generative AI models (like LLMs) and Agentic AI applications without relying on the cloud. Run models with up to 200 billion parameters on your desk using 128 GB of unified memory (shared across CPU, GPU, and NPU). This eliminates cloud costs, latency, data privacy concerns, and internet dependency. ~This is for AI developers, researchers, engineers, and creators. ~Teams or individuals wanting faster iteration, data privacy, and cost savings compared to cloud APIs. ~Users experimenting with open-source models, fine-tuning, inference, and production-like local deployment Source: AMD CES 2026

Mike

16,097 次观看 • 4 个月前

$WDC CEO says storage demand and requirement compound and grow whether AI runs on US frontier models or cheaper open-source ones. "Whether it's the traditional frontier models that you've seen in the U.S. that are premium-based, or some of the new open-source, more economical models, the underlying requirement is still that they're going to require a lot of data to support the training of these models." "These models are going to generate a lot more data that requires storage for us." "So we view it as very positive, because even as compute maybe drives greater efficiency in terms of compute and memory resources, the requirement for storage is just going to compound and grow." "So we view it very positively for demand going forward."

$WDC CEO says storage demand and requirement compound and grow whether AI runs on US frontier models or cheaper open-source ones. "Whether it's the traditional frontier models that you've seen in the U.S. that are premium-based, or some of the new open-source, more economical models, the underlying requirement is still that they're going to require a lot of data to support the training of these models." "These models are going to generate a lot more data that requires storage for us." "So we view it as very positive, because even as compute maybe drives greater efficiency in terms of compute and memory resources, the requirement for storage is just going to compound and grow." "So we view it very positively for demand going forward."

Fireside Alpha

14,114 次观看 • 1 个月前

Jensen Huang on today's All-In Summit: open-source AI is what allows an entire economy to compete. "in the last six months, $400 billion of venture funding went into AI-native companies. Eighty percent of them use open models. If not for open models, how could they build their dream? The vast majority of the world’s contribution to open source today is coming from China. But once you download it, it’s yours. We fork it. We improve it. We make it ours." ---- Full video on "All-In Podcast" YouTube channel (link in comment)

Jensen Huang on today's All-In Summit: open-source AI is what allows an entire economy to compete. "in the last six months, $400 billion of venture funding went into AI-native companies. Eighty percent of them use open models. If not for open models, how could they build their dream? The vast majority of the world’s contribution to open source today is coming from China. But once you download it, it’s yours. We fork it. We improve it. We make it ours." ---- Full video on "All-In Podcast" YouTube channel (link in comment)

Rohan Paul

32,272 次观看 • 13 天前

Today, we’re thrilled to unveil the Arcee.ai Foundation Models, a new family of GenAI models designed from the ground up for enterprise reality. The first release—AFM-4.5B—is a 4.5-billion-parameter frontier model that delivers excellent accuracy, strict compliance, and very high cost efficiency. In short: enterprise-grade intelligence that can run anywhere—on a smartphone, at the edge, or in the cloud. For a quick taste, you can test AFM-4.5B in our playground and on For a deeper dive into the model’s training pipeline and benchmarks, details are available in our technical blog post. ➡️ Arcee AI playground: ➡️ Together AI playground: ➡️ Launch blog post: ➡️ Technical blog post: PS: As we’ll now focus on our AFM foundation models, and because we love open-source, we're opening up access to our previously closed-source language models. Details in the tech blog post!

Today, we’re thrilled to unveil the Arcee.ai Foundation Models, a new family of GenAI models designed from the ground up for enterprise reality. The first release—AFM-4.5B—is a 4.5-billion-parameter frontier model that delivers excellent accuracy, strict compliance, and very high cost efficiency. In short: enterprise-grade intelligence that can run anywhere—on a smartphone, at the edge, or in the cloud. For a quick taste, you can test AFM-4.5B in our playground and on For a deeper dive into the model’s training pipeline and benchmarks, details are available in our technical blog post. ➡️ Arcee AI playground: ➡️ Together AI playground: ➡️ Launch blog post: ➡️ Technical blog post: PS: As we’ll now focus on our AFM foundation models, and because we love open-source, we're opening up access to our previously closed-source language models. Details in the tech blog post!

Arcee.ai

21,934 次观看 • 1 年前

>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 次观看 • 2 个月前

Announcing the Open Model Series: a new series on open source AI models, with merve of Hugging Face. Hugging Face is the open model hub: it hosts model weights, datasets, evals, & demo apps, and maintains the open-source libraries (like transformers) used to run and train them. I think of it like GitHub, but for AI models. You can download open models and run them on your own machines. In this series we'll cover what open source and open weight models are, how they work, how to run them yourself, and why they're important. By the way - this video is edited entirely with Claude! I can make a tutorial on how I did it. If there's a question you want answered about open models, leave it in the comments. We will read them all. Episode 1 is out soon.

Announcing the Open Model Series: a new series on open source AI models, with merve of Hugging Face. Hugging Face is the open model hub: it hosts model weights, datasets, evals, & demo apps, and maintains the open-source libraries (like transformers) used to run and train them. I think of it like GitHub, but for AI models. You can download open models and run them on your own machines. In this series we'll cover what open source and open weight models are, how they work, how to run them yourself, and why they're important. By the way - this video is edited entirely with Claude! I can make a tutorial on how I did it. If there's a question you want answered about open models, leave it in the comments. We will read them all. Episode 1 is out soon.

Harper Carroll

30,392 次观看 • 1 个月前

NVIDIA just released a new open source transcription model, Nemotron Speech ASR, designed from the ground up for low-latency use cases like voice agents. Here's a voice agent built with this new model. 24ms transcription finalization and total voice-to-voice inference time under 500ms. This agent actually uses *three* NVIDIA open source models: - Nemotron Speech ASR - Nemotron 3 Nano 30GB in a 4-bit quant (released in December) - A preview checkpoint of the upcoming Magpie text-to-speech model These models are all truly open source: weights, training data, training code, and inference code. This is a big deal! Jensen said in the CES keynote yesterday that he expects open source models to catch up to proprietary models this year in a number of categories. NVIDIA is putting their weight behind making this happen. (As Alan Kay said, the best way to predict the future is to invent it.) The code for this agent is open source too, of course. You can deploy it to production with Modal and Pipecat AI cloud, or run locally on an NVIDIA DGX Spark or RTX 5090.

NVIDIA just released a new open source transcription model, Nemotron Speech ASR, designed from the ground up for low-latency use cases like voice agents. Here's a voice agent built with this new model. 24ms transcription finalization and total voice-to-voice inference time under 500ms. This agent actually uses three NVIDIA open source models: - Nemotron Speech ASR - Nemotron 3 Nano 30GB in a 4-bit quant (released in December) - A preview checkpoint of the upcoming Magpie text-to-speech model These models are all truly open source: weights, training data, training code, and inference code. This is a big deal! Jensen said in the CES keynote yesterday that he expects open source models to catch up to proprietary models this year in a number of categories. NVIDIA is putting their weight behind making this happen. (As Alan Kay said, the best way to predict the future is to invent it.) The code for this agent is open source too, of course. You can deploy it to production with Modal and Pipecat AI cloud, or run locally on an NVIDIA DGX Spark or RTX 5090.

kwindla

274,867 次观看 • 8 个月前

Introducing Silico: the platform for building AI models with the precision of written software. Silico lets researchers and engineers see inside their models, debug failures, and intentionally design them from the ground up. Early access is open now. 🧵(1/10)

Introducing Silico: the platform for building AI models with the precision of written software. Silico lets researchers and engineers see inside their models, debug failures, and intentionally design them from the ground up. Early access is open now. 🧵(1/10)

Goodfire

112,156 次观看 • 5 个月前