Все опубликованные аналитические материалы, исследования суверенного стека, новости экосистемы и обзоры моделей ИИ.
Transformers.js brings state-of-the-art machine learning to JavaScript. You can think of it as a library for running AI models locally in the browser through a single high-level API that handles model loading, pre-processing, inference, and post-processing. Hi, I'm Nico and in…
Good morning everyone. How's it going today? Today we're going to be talking about the fine web data set. Now, this is a project by Hugging Face and it is a great example of how to gather all the data that you need to train a large language model. This actually is also available…
Год назад тысячи людей по всему миру потратили один уик-энд на создание роботов, которых они никогда раньше не создавали. Вот что они сделали и где они сейчас находятся. >> Привет, я Никодим из Hugging Face. Ровно год назад хакатон Robot Worldwide собрал тысячи…
Today we are going to hear about the shift towards agentic engineering and increased code output. Welcome Paul. >> One minute into this conference Jonathan said Let's talk about the future of software engineering. Let's have a conversation. Let's figure it out. That's exactly…
This is Leloup, a graphical user interface for the Robot Library. It lets you teleoperate the robots, collect the data set, train models both on your own hardware and with powerful GPUs available through hugging face. And finally, you can of course deploy your working policies…
Uh Andrew here from ZenCode. So today we're going to talk about some very interesting results from our internal lab team that we first kind of tested on different benchmark and evals and then on our own team being guinea pigs. We we got about 50 engineers and then happily…
I'm going to talk to you a little bit about what does it really mean for you to expand from just building one or two agents for your company to building hundreds, you know, the next order of magnitude. And I'll be telling you a little bit about what we've learned at DataDog…
Good morning everyone. How's it going today? Uh today we're going to be taking a look at the Hermes agent and its architecture. So, we're going to be covering a pretty high-level overview of how this agent is built so that you understand a little bit better how to use it and…
Good morning, everyone. How's it going? Today, we're going to be introducing Tao, which is a coding agent that is designed to be educational and to help you build your own coding agents. So, the objective of this project is not to make it compete with established agents like Pi…
This paper is doing something if I understand correctly that is very similar to what pipeline RL proposed for reinforcement learning which is where you want to basically have like really decoupled generators from learners from trainers um but you have the additional complexity…
We're trying to optimize that uh vicious cycle of accuracy, cost, latency, uh and you know, it always feels like you can't really have all three. Uh you optimize for one, add another one, the the last one drops. So, that would be what I will be talking about today. How can we…
Всем привет. Если вы работаете с ИИ, у вас наверняка уже есть аккаунт на Hugging Face Hub, где удобно демонстрировать свои модели, скачивать модели с открытым кодом, использовать публичные наборы данных и создавать демоверсии своих моделей. Но публичный хаб—это только начало. В…
I'm here today to talk about a new frontier that we see opening in AI. That frontier is collaborative AI systems. Um that frontier is going to unlock a new category of systems to be built. Systems that were not possible before. But before I do that, I just want to say a few…
So, Enterprise AI is changing and it's changing very fast. Enterprises are moving faster and faster from generic intelligence that they rent to specialized intelligence that they own. When the whole generative AI wave started in November of 2022 when ChatGPT came out, then 2023…
Let [clears throat] me think where to start. Okay, so we've got the Kim K3 report and it's quite a beast in terms of details, which is pretty awesome. And the bits that I focused on were really just the the pre- uh sorry, the post-training parts. And even that I wasn't able to…
Hi everybody. My name is Luke Kim. I'm the founder and CEO of Spice AI. We are a data and AI infrastructure startup out of Bellevue. We build an open source project which is a data platform for AI context which I will show you soon. So if you think about the SAS era the last 10…
So, recently it has become very clear that coding harnesses are one of the most important parts when creating reliable AI workflows. And what we're going to be doing today is we're going to be taking a look at one coding harness that we're developing internally right here called…
Uh, this paper is actually quite interesting. It dives into something that that has been I guess a question an open question for the Ember communities the Um, we can call them like scaling laws for post-training. And then how is pre-training allocation of compute impact…
So, uh let me start with uh one thing that there's so much to unpack on this talk and we don't have enough time. So, we're going to be rushing for a hot second, but it's going to be fun. And the idea here is I I tried to fit the entire title of this talk in one slide and I…
So, I'm Andy. I'm the CTO and co-founder of Verifai. Um I'm going to talk about simulation environments for agents. Now, I'm assuming you are here because you're very interested in agents, but I also do want to see a show of hands, who here is building AI agents, not just using…
So, let's suppose that you're working with your coding agent and you already have accumulated 50k tokens of context. And then you send an additional 1k tokens for an additional query, an additional question that you send to it. So, that adds a full context of 51,000 tokens. Are…
I'm Era. I'm going to be talking about emails, specifically AI emails like coding agent emails and stuff. And I'm going to talk about how they're broken and how you could still use them anyway. Uh before I start, I just I want to say one thing. It's just like it boggles my mind…
Okay, so this is quite um interesting paper and a bit uh surprising that uh how it works, but uh basically the idea is that uh in post-training, you have usually uh a pipeline where you have a model and then you uh run RL over that model. And ideally that uh post-RL model should…
Good morning everyone. If you want your agent to remember things across conversations about yourself, about itself, and what it learns about the world, you're going to need an agent memory system. And today, that's what we're going to be explaining. First of all, we're going to…
Hi, I'm Andrew Ng. I want to show you how to build fun, useful software using AI, regardless of your background or training. If you're intrigued about using AI or live coding to build cool software, but aren't sure where to start, this course is for you. Let me show you a funny…
I'm thrilled to introduce this course on fast and efficient LLM inference, built in partnership with Red Hat, and taught by Sergey Kliger. Deploying open source LLMs efficiently, so they can serve many users at once with low latency and reasonable cost, is challenging. In this…
All right. So this paper um is about a model called Faraday um which was trained by this new startup called inherent and what they train the model to do is to do a very narrow part of like scientific replication. So if I understood correctly, basically they want to take a bunch…
Yeah, I think it's a really cool paper. Um, I think we've talked a bit about this before or like I think it has been a bit in the in the common discourse that like now that implementation is sped up so much, um, like the the value that researchers can provide is more in their…
Cerebras, which I've been a fan of for a long time, is one of the leaders of fast inference. I'm excited to introduce this short course on fast LM inference built in partnership with Cerebras and taught by Jem Weigall, Satya, and Sara Hooker. Fast inference make new classes of…
We just shipped more than 200 WebGPU kernels for browser AI and they are all available open source on the Hugging Face Hub. We also shipped a new library called Hugging Face Kernels that allows you to load and to execute all of those kernels. Let's have a look at this little…