Video wird geladen...

Video konnte nicht geladen werden

Zur Startseite

Claude distillation has been a big topic this week while I am (coincidentally) writing Chapter 8 on model distillation. In that context, I shared some utilities to generate distillation data from all sorts of open-weight models via OpenRouter and Ollama:

rasbt's profile picture

Sebastian Raschka

474,171 subscribers

62,838 Aufrufe • vor 4 Monaten •via X (Twitter)

BildungNachrichten & PolitikWissenschaft & Technologie
LIVE

Anya Rossi• Live Now

Private livecam show

0 Kommentare

Keine Kommentare verfügbar

Kommentare vom Original-Post werden hier angezeigt

Ähnliche Videos

New lecture for the book! Nominally about synthetic data, but mostly is a walk through of the distillation literature from the Hinton 2015 paper to multi-teach on-policy distillation of today! At 7.4 hours of video in my post-training brain dump and counting :) It was fun to stare at the math long enough and talk through the 3-4 core changes that needed to be made to the original formulation to have on-policy distillation be ready for the mainstream like it is today (and in RL frameworks). Otherwise, I include a bit of a history lesson for how synthetic data generally slowly took over all post-training data research (it wasn't always the case)! Then I do some 101 review on constitutional AI, rubrics, and other popular methods. 00:00 The emergence of synthetic data 10:50 Background on teacher-student knowledge-distillation 24:47: On-policy distillation (OPD, MOPD, and OPSD) 37:11 Constitutional AI & AI Feedback 45:50 Rubrics as rewards & conclusions Ofc, watch on YouTube etc.
49:40

New lecture for the book! Nominally about synthetic data, but mostly is a walk through of the distillation literature from the Hinton 2015 paper to multi-teach on-policy distillation of today! At 7.4 hours of video in my post-training brain dump and counting :) It was fun to stare at the math long enough and talk through the 3-4 core changes that needed to be made to the original formulation to have on-policy distillation be ready for the mainstream like it is today (and in RL frameworks). Otherwise, I include a bit of a history lesson for how synthetic data generally slowly took over all post-training data research (it wasn't always the case)! Then I do some 101 review on constitutional AI, rubrics, and other popular methods. 00:00 The emergence of synthetic data 10:50 Background on teacher-student knowledge-distillation 24:47: On-policy distillation (OPD, MOPD, and OPSD) 37:11 Constitutional AI & AI Feedback 45:50 Rubrics as rewards & conclusions Ofc, watch on YouTube etc.

Nathan Lambert

91,890 Aufrufe • vor 25 Tagen

Presto! Distilling Steps and Layers for Accelerating Music Generation Despite advances in diffusion-based text-to-music (TTM) methods, efficient, high-quality generation remains a challenge. We introduce Presto!, an approach to inference acceleration for score-based diffusion transformers via reducing both sampling steps and cost per step. To reduce steps, we develop a new score-based distribution matching distillation (DMD) method for the EDM-family of diffusion models, the first GAN-based distillation method for TTM. To reduce the cost per step, we develop a simple, but powerful improvement to a recent layer distillation method that improves learning via better preserving hidden state variance. Finally, we combine our step and layer distillation methods together for a dual-faceted approach. We evaluate our step and layer distillation methods independently and show each yield best-in-class performance. Our combined distillation method can generate high-quality outputs with improved diversity, accelerating our base model by 10-18x (230/435ms latency for 32 second mono/stereo 44.1kHz, 15x faster than comparable SOTA) -- the fastest high-quality TTM to our knowledge.
1:20

Presto! Distilling Steps and Layers for Accelerating Music Generation Despite advances in diffusion-based text-to-music (TTM) methods, efficient, high-quality generation remains a challenge. We introduce Presto!, an approach to inference acceleration for score-based diffusion transformers via reducing both sampling steps and cost per step. To reduce steps, we develop a new score-based distribution matching distillation (DMD) method for the EDM-family of diffusion models, the first GAN-based distillation method for TTM. To reduce the cost per step, we develop a simple, but powerful improvement to a recent layer distillation method that improves learning via better preserving hidden state variance. Finally, we combine our step and layer distillation methods together for a dual-faceted approach. We evaluate our step and layer distillation methods independently and show each yield best-in-class performance. Our combined distillation method can generate high-quality outputs with improved diversity, accelerating our base model by 10-18x (230/435ms latency for 32 second mono/stereo 44.1kHz, 15x faster than comparable SOTA) -- the fastest high-quality TTM to our knowledge.

AK

30,430 Aufrufe • vor 1 Jahr

I had an awesome time interviewing idan shenfeld and Jonas Hübotter from MIT and ETH Zurich about self-distillation. this very promising post-training paradigm where the model acts as its own teacher by conditioning on environment feedback or demonstrations. we cover the SDPO algo for reinforcement learning with rich feedback and SDFT for continual learning without forgetting along with many applications. we dig into how it works, why it's simpler and faster than GRPO, and where this is already showing up in production systems. table of content: 0:00 - what is self distillation 2:50 - idan (MIT) and jonas (ETH Zurich) introduction and motivation 18:40 - different perspective of on-policy self-distillation (presentation) 36:00 - metacognition and specificity in self-distillation 37:24 - very long hard task and self-distillation 42:00 - continual learning with self-distillation (presentation) 1:16:50 - what is next in this research direction? 1:20:00 - is there any experience with subjective feedbacks? 1:22:50 - quality vs number of feedbacks? 1:26:40 - what setting would self-distillation struggle vs GRPO? the slides were super crisp really cool of them to share! enjoy my guys 🌹
1:31:19

I had an awesome time interviewing idan shenfeld and Jonas Hübotter from MIT and ETH Zurich about self-distillation. this very promising post-training paradigm where the model acts as its own teacher by conditioning on environment feedback or demonstrations. we cover the SDPO algo for reinforcement learning with rich feedback and SDFT for continual learning without forgetting along with many applications. we dig into how it works, why it's simpler and faster than GRPO, and where this is already showing up in production systems. table of content: 0:00 - what is self distillation 2:50 - idan (MIT) and jonas (ETH Zurich) introduction and motivation 18:40 - different perspective of on-policy self-distillation (presentation) 36:00 - metacognition and specificity in self-distillation 37:24 - very long hard task and self-distillation 42:00 - continual learning with self-distillation (presentation) 1:16:50 - what is next in this research direction? 1:20:00 - is there any experience with subjective feedbacks? 1:22:50 - quality vs number of feedbacks? 1:26:40 - what setting would self-distillation struggle vs GRPO? the slides were super crisp really cool of them to share! enjoy my guys 🌹

Yacine Mahdid

13,023 Aufrufe • vor 2 Monaten

Meta presents Video Editing via Factorized Diffusion Distillation We introduce Emu Video Edit (EVE), a model that establishes a new state-of-the art in video editing without relying on any supervised video editing data. To develop EVE we separately train an image editing
0:24

Meta presents Video Editing via Factorized Diffusion Distillation We introduce Emu Video Edit (EVE), a model that establishes a new state-of-the art in video editing without relying on any supervised video editing data. To develop EVE we separately train an image editing

AK

115,598 Aufrufe • vor 2 Jahren

ProlificDreamer: High-Fidelity and Diverse Text-to-3D Generation with Variational Score Distillation propose to model the 3D parameter as a random variable instead of a constant as in SDS and present variational score distillation (VSD), a principled particle-based variational framework to explain and address the aforementioned issues in text-to-3D generation. We show that SDS is a special case of VSD and leads to poor samples with both small and large CFG weights. In comparison, VSD works well with various CFG weights as ancestral sampling from diffusion models and simultaneously improves the diversity and sample quality with a common CFG weight (i.e., 7.5). We further present various improvements in the design space for text-to-3D such as distillation time schedule and density initialization, which are orthogonal to the distillation algorithm yet not well explored. Our overall approach, dubbed ProlificDreamer, can generate high rendering resolution (i.e., 512times512) and high-fidelity NeRF with rich structure and complex effects (e.g., smoke and drops). Further, initialized from NeRF, meshes fine-tuned by VSD are meticulously detailed and photo-realistic paper page:
0:43

ProlificDreamer: High-Fidelity and Diverse Text-to-3D Generation with Variational Score Distillation propose to model the 3D parameter as a random variable instead of a constant as in SDS and present variational score distillation (VSD), a principled particle-based variational framework to explain and address the aforementioned issues in text-to-3D generation. We show that SDS is a special case of VSD and leads to poor samples with both small and large CFG weights. In comparison, VSD works well with various CFG weights as ancestral sampling from diffusion models and simultaneously improves the diversity and sample quality with a common CFG weight (i.e., 7.5). We further present various improvements in the design space for text-to-3D such as distillation time schedule and density initialization, which are orthogonal to the distillation algorithm yet not well explored. Our overall approach, dubbed ProlificDreamer, can generate high rendering resolution (i.e., 512times512) and high-fidelity NeRF with rich structure and complex effects (e.g., smoke and drops). Further, initialized from NeRF, meshes fine-tuned by VSD are meticulously detailed and photo-realistic paper page:

AK

46,151 Aufrufe • vor 3 Jahren

DeepSeek Didn’t Catch Up — It Cheated. China’s AI “Breakthrough” Is Built on Stolen American Tech. In this video, I break down explosive allegations that DeepSeek used banned U.S. chips and illicit model distillation to train its newest system. If true, this is IP extraction wrapped in marketing. We examine what distillation really means, why export controls matter, and how dependent China’s AI ecosystem still is on American hardware and research.
11:22

DeepSeek Didn’t Catch Up — It Cheated. China’s AI “Breakthrough” Is Built on Stolen American Tech. In this video, I break down explosive allegations that DeepSeek used banned U.S. chips and illicit model distillation to train its newest system. If true, this is IP extraction wrapped in marketing. We examine what distillation really means, why export controls matter, and how dependent China’s AI ecosystem still is on American hardware and research.

Ken Cao-The China Crash Chronicle

12,668 Aufrufe • vor 4 Monaten

What happens to Chinese AI once American companies stop releasing frontier models internationally? Gavin Baker says that China's made a "terrible mistake" by relying on its own chips and distilling American models: "It feels like the administration has been willing to let them buy H200s or B30s. And I think [the Chinese] have this crazy belief that, 'Oh, our own internal chips are good enough.' They're not." "And what's making them think that is that Chinese labs are very, very good at distillation...we've all seen those iPhone farms in China. You know they've got the same thing for distillation. They've got 100,000 endpoints distilling these models across every API available." "They've gotten really, really good at that. But all that goes away if people stop releasing these models at the frontier. And I think Mythos is a sign of things to come there."
1:21

What happens to Chinese AI once American companies stop releasing frontier models internationally? Gavin Baker says that China's made a "terrible mistake" by relying on its own chips and distilling American models: "It feels like the administration has been willing to let them buy H200s or B30s. And I think [the Chinese] have this crazy belief that, 'Oh, our own internal chips are good enough.' They're not." "And what's making them think that is that Chinese labs are very, very good at distillation...we've all seen those iPhone farms in China. You know they've got the same thing for distillation. They've got 100,000 endpoints distilling these models across every API available." "They've gotten really, really good at that. But all that goes away if people stop releasing these models at the frontier. And I think Mythos is a sign of things to come there."

TBPN

122,701 Aufrufe • vor 1 Monat

What happens when robot world models learn from human experience at scale? 🤔 DreamDojo from NVIDIA Research is a generalist robot world model pretrained on 44K hours of egocentric human videos and then post-trained on robot data to generalize across new objects and environments. After distillation, it runs at 10 FPS for live teleoperation, policy evaluation, and model-based planning. Read the ICML paper to learn more 📄
0:32

What happens when robot world models learn from human experience at scale? 🤔 DreamDojo from NVIDIA Research is a generalist robot world model pretrained on 44K hours of egocentric human videos and then post-trained on robot data to generalize across new objects and environments. After distillation, it runs at 10 FPS for live teleoperation, policy evaluation, and model-based planning. Read the ICML paper to learn more 📄

NVIDIA Robotics

21,506 Aufrufe • vor 12 Tagen

🆕 How to run (and finetune) open source AI models with a simple API! In 5 mins, I go over how to: ◆ Generate text with DeepSeek R1 & Llama 3 ◆ Generate code with Qwen on LlamaCoder ◆ Generate images with Flux on BlinkShot ◆ Finetune a model on your own data & run it
4:44

🆕 How to run (and finetune) open source AI models with a simple API! In 5 mins, I go over how to: ◆ Generate text with DeepSeek R1 & Llama 3 ◆ Generate code with Qwen on LlamaCoder ◆ Generate images with Flux on BlinkShot ◆ Finetune a model on your own data & run it

Hassan

30,236 Aufrufe • vor 1 Jahr

Here's my conversation with Lucas Atkins and the team at Arcee.ai on their path to training and releasing Trinity Large today. From going all in on open models built end to end in the US 6 months ago to having the model in hand is no easy feet. I loved this conversation on how to design a startup around open models and take a bold step to scale it up. I'm openly an Arcee fan, watching them take risk and pull it off. We discuss: - The state (and future) of open vs. closed models, - The business of selling open models for on-prem deployments, - The story of Arcee AI & going “all-in” on this training run, - The ATOM project, - Building frontier model training teams in 6 months, - and other great topics. I really loved this one, and think you well too. Chapters: 00:00:00 Intro: Arcee AI, Trinity Models & Trinity Large 00:08:26 Transitioning a Company to Pre-training 00:13:00 Technical Decisions: Muon and MoE 00:18:41 Scaling and MoE Training Pain 00:23:14 Post-training and RL Strategies 00:28:09 Team Structure and Data Scaling 00:31:31 The Trinity Manifesto: US Open Weights 00:42:31 Specialized Models and Distillation 00:47:12 Infrastructure and Hosting 400B 00:50:53 Open Source as a Business Moat 00:56:31 Predictions: Best Model in 2026 01:02:29 Lightning Round & Conclusions More great open model builder podcasts coming soon!
1:12:15

Here's my conversation with Lucas Atkins and the team at Arcee.ai on their path to training and releasing Trinity Large today. From going all in on open models built end to end in the US 6 months ago to having the model in hand is no easy feet. I loved this conversation on how to design a startup around open models and take a bold step to scale it up. I'm openly an Arcee fan, watching them take risk and pull it off. We discuss: - The state (and future) of open vs. closed models, - The business of selling open models for on-prem deployments, - The story of Arcee AI & going “all-in” on this training run, - The ATOM project, - Building frontier model training teams in 6 months, - and other great topics. I really loved this one, and think you well too. Chapters: 00:00:00 Intro: Arcee AI, Trinity Models & Trinity Large 00:08:26 Transitioning a Company to Pre-training 00:13:00 Technical Decisions: Muon and MoE 00:18:41 Scaling and MoE Training Pain 00:23:14 Post-training and RL Strategies 00:28:09 Team Structure and Data Scaling 00:31:31 The Trinity Manifesto: US Open Weights 00:42:31 Specialized Models and Distillation 00:47:12 Infrastructure and Hosting 400B 00:50:53 Open Source as a Business Moat 00:56:31 Predictions: Best Model in 2026 01:02:29 Lightning Round & Conclusions More great open model builder podcasts coming soon!

Nathan Lambert

26,872 Aufrufe • vor 5 Monaten

Introducing Ollama fully Integrated into VSCode with CodeGPT Now you can download open-source models from ollama with a single click and install them as code assistants in VSCode. In this case, I download the qwen2:0.5b (Qwen) model in under a minute, and it becomes fully functional with just one click. The era of locally running code assistants has begun! 🚀
1:14

Introducing Ollama fully Integrated into VSCode with CodeGPT Now you can download open-source models from ollama with a single click and install them as code assistants in VSCode. In this case, I download the qwen2:0.5b (Qwen) model in under a minute, and it becomes fully functional with just one click. The era of locally running code assistants has begun! 🚀

Daniel San

101,567 Aufrufe • vor 2 Jahren

👀 Elon Musk Says Grokipedia Will Become a Backup of All Knowledge for Future Civilizations “The idea behind this Galactica is to create an open source … distillation of all knowledge … And then we want to create copies of this and distribute these copies throughout Earth, and even put them on the moon and Mars and out of deep space … So in the worst case scenario, future civilization can see what we learned and maybe pick things up from there.” 📸 Elon Musk
1:33

👀 Elon Musk Says Grokipedia Will Become a Backup of All Knowledge for Future Civilizations “The idea behind this Galactica is to create an open source … distillation of all knowledge … And then we want to create copies of this and distribute these copies throughout Earth, and even put them on the moon and Mars and out of deep space … So in the worst case scenario, future civilization can see what we learned and maybe pick things up from there.” 📸 Elon Musk

Chief Nerd

61,195 Aufrufe • vor 7 Monaten

“FOR I AM DEATH” came out a week ago today and I just wanted to thank you all for such an amazing response 🔥 I was literally on a plane flying home from AC/DC when it came out and it’s been a crazy ride all week thanks to all of you! ❤️ This is just the beginning, some big surprises are coming, stay tuned…xxT
0:35

“FOR I AM DEATH” came out a week ago today and I just wanted to thank you all for such an amazing response 🔥 I was literally on a plane flying home from AC/DC when it came out and it’s been a crazy ride all week thanks to all of you! ❤️ This is just the beginning, some big surprises are coming, stay tuned…xxT

The Pretty Reckless

16,616 Aufrufe • vor 10 Monaten

I've been using Claude Code for non-coding tasks, like sorting emails and analyzing my spending This was annoying to setup because Claude doesn’t have the right context - it can’t see my email or my credit card statements So we built a virtual filesystem that turns all of my data sources (gmail, notion, gdrive and more) into local folders on my computer I can just describe what I need (e.g. "invoices from the last week") and a folder appears on my computer with exactly those files This gives Claude context, but it also gives it memory - since it has a local filesystem that syncs in the background, it continuously gets access to my data The more things you expose as files, the more Claude can do for you
2:50

I've been using Claude Code for non-coding tasks, like sorting emails and analyzing my spending This was annoying to setup because Claude doesn’t have the right context - it can’t see my email or my credit card statements So we built a virtual filesystem that turns all of my data sources (gmail, notion, gdrive and more) into local folders on my computer I can just describe what I need (e.g. "invoices from the last week") and a folder appears on my computer with exactly those files This gives Claude context, but it also gives it memory - since it has a local filesystem that syncs in the background, it continuously gets access to my data The more things you expose as files, the more Claude can do for you

Eli Mernit

67,006 Aufrufe • vor 5 Monaten

I am proud of what 2 weeks of no sugar and less carbs has done. This week has been a fasting week. It has been very challenging, I have lost a lot of body fat and still working on it. I have never believed that age is to define your quality of life
0:40

I am proud of what 2 weeks of no sugar and less carbs has done. This week has been a fasting week. It has been very challenging, I have lost a lot of body fat and still working on it. I have never believed that age is to define your quality of life

Edgar Gaines

40,271 Aufrufe • vor 2 Jahren

DreamCraft3D: Hierarchical 3D Generation with Bootstrapped Diffusion Prior paper page: present DreamCraft3D, a hierarchical 3D content generation method that produces high-fidelity and coherent 3D objects. We tackle the problem by leveraging a 2D reference image to guide the stages of geometry sculpting and texture boosting. A central focus of this work is to address the consistency issue that existing works encounter. To sculpt geometries that render coherently, we perform score distillation sampling via a view-dependent diffusion model. This 3D prior, alongside several training strategies, prioritizes the geometry consistency but compromises the texture fidelity. We further propose Bootstrapped Score Distillation to specifically boost the texture. We train a personalized diffusion model, Dreambooth, on the augmented renderings of the scene, imbuing it with 3D knowledge of the scene being optimized. The score distillation from this 3D-aware diffusion prior provides view-consistent guidance for the scene. Notably, through an alternating optimization of the diffusion prior and 3D scene representation, we achieve mutually reinforcing improvements: the optimized 3D scene aids in training the scene-specific diffusion model, which offers increasingly view-consistent guidance for 3D optimization. The optimization is thus bootstrapped and leads to substantial texture boosting. With tailored 3D priors throughout the hierarchical generation, DreamCraft3D generates coherent 3D objects with photorealistic renderings, advancing the state-of-the-art in 3D content generation.
2:32

DreamCraft3D: Hierarchical 3D Generation with Bootstrapped Diffusion Prior paper page: present DreamCraft3D, a hierarchical 3D content generation method that produces high-fidelity and coherent 3D objects. We tackle the problem by leveraging a 2D reference image to guide the stages of geometry sculpting and texture boosting. A central focus of this work is to address the consistency issue that existing works encounter. To sculpt geometries that render coherently, we perform score distillation sampling via a view-dependent diffusion model. This 3D prior, alongside several training strategies, prioritizes the geometry consistency but compromises the texture fidelity. We further propose Bootstrapped Score Distillation to specifically boost the texture. We train a personalized diffusion model, Dreambooth, on the augmented renderings of the scene, imbuing it with 3D knowledge of the scene being optimized. The score distillation from this 3D-aware diffusion prior provides view-consistent guidance for the scene. Notably, through an alternating optimization of the diffusion prior and 3D scene representation, we achieve mutually reinforcing improvements: the optimized 3D scene aids in training the scene-specific diffusion model, which offers increasingly view-consistent guidance for 3D optimization. The optimization is thus bootstrapped and leads to substantial texture boosting. With tailored 3D priors throughout the hierarchical generation, DreamCraft3D generates coherent 3D objects with photorealistic renderings, advancing the state-of-the-art in 3D content generation.

AK

161,530 Aufrufe • vor 2 Jahren

“The Agenda is a frightening watch. But I tell you now, watching it and knowing its contents is the vital first step on the road to recovery for populations addicted to the drug that has been used to make us compliant to all that's been happening - convenience.” "Having seen The Agenda, you have the distillation, the executive summary of frankly all a person needs to know about the threat we face as a species." "The man-made catastrophe that most threatens us has a name. And its name is Agenda 2030." High praise from Neil Oliver - thank you 🙏 Link to source 🧵
9:55

“The Agenda is a frightening watch. But I tell you now, watching it and knowing its contents is the vital first step on the road to recovery for populations addicted to the drug that has been used to make us compliant to all that's been happening - convenience.” "Having seen The Agenda, you have the distillation, the executive summary of frankly all a person needs to know about the threat we face as a species." "The man-made catastrophe that most threatens us has a name. And its name is Agenda 2030." High praise from Neil Oliver - thank you 🙏 Link to source 🧵

Oracle Films

47,972 Aufrufe • vor 1 Jahr

China just released an open source AI model that matches the best closed models from OpenAI and Anthropic. Gavin Baker explained exactly how they did it and the answer should concern every American AI lab. The model is called GLM 5.2. It was built by Z. AI. You get 744 billion parameters, 1 million token context window and its MIT license, meaning anyone can download it, fork it, build a company on it, with no restrictions and no Dario. It scored 51 points on the artificial analysis intelligence index. The highest score any open weight model has ever achieved. It beat GPT 5.5 on the frontier software engineering benchmark. It trails Claude Opus 4.8 by less than one percentage point. And it costs 85% less to run than GPT 5.5 for comparable performance. Gavin Baker said on the All-In podcast that this model has challenged some of his beliefs. Then he explained how China built it. The method is called distillation. Just think of tens of thousands of phones and computers running simultaneously, all hitting the frontier model APIs through masked accounts, asking specific questions, and harvesting what happens inside the model when it answers. Every reasoning step, every token. The entire thinking process gets recorded and fed back into the Chinese model during training. It is a cheat sheet. It is the answer key to the exam. And here is the part that should worry everyone. Sacks said it plainly. China was already nine months behind American models. But now that GLM 5.2 is good enough to run its own reinforcement learning, it can improve itself without needing to distill from American models anymore. The cheat sheet let them get close enough to start writing their own answers. Sacks said we are six months behind on the model and 24 months behind on silicon and they are only a few months behind in total. The Z. AI founder told Elon Musk directly that open weight fable-level capability will be here before Q1 2027. Every restriction Anthropic lobbied for, every self-imposed safety guardrail, every month of delay in releasing American frontier models accelerated this. The Chinese labs were not under those restrictions. They were not going to wait. The composable model future Gavin described, where every enterprise runs a frontier model alongside their own fine-tuned open weight model, is coming regardless of what American labs do next. The question is just whether the open weight half of that stack is American or Chinese. Right now it is Chinese. WATCH THE FULL PODCAST ON The All-In Podcast
3:21

China just released an open source AI model that matches the best closed models from OpenAI and Anthropic. Gavin Baker explained exactly how they did it and the answer should concern every American AI lab. The model is called GLM 5.2. It was built by Z. AI. You get 744 billion parameters, 1 million token context window and its MIT license, meaning anyone can download it, fork it, build a company on it, with no restrictions and no Dario. It scored 51 points on the artificial analysis intelligence index. The highest score any open weight model has ever achieved. It beat GPT 5.5 on the frontier software engineering benchmark. It trails Claude Opus 4.8 by less than one percentage point. And it costs 85% less to run than GPT 5.5 for comparable performance. Gavin Baker said on the All-In podcast that this model has challenged some of his beliefs. Then he explained how China built it. The method is called distillation. Just think of tens of thousands of phones and computers running simultaneously, all hitting the frontier model APIs through masked accounts, asking specific questions, and harvesting what happens inside the model when it answers. Every reasoning step, every token. The entire thinking process gets recorded and fed back into the Chinese model during training. It is a cheat sheet. It is the answer key to the exam. And here is the part that should worry everyone. Sacks said it plainly. China was already nine months behind American models. But now that GLM 5.2 is good enough to run its own reinforcement learning, it can improve itself without needing to distill from American models anymore. The cheat sheet let them get close enough to start writing their own answers. Sacks said we are six months behind on the model and 24 months behind on silicon and they are only a few months behind in total. The Z. AI founder told Elon Musk directly that open weight fable-level capability will be here before Q1 2027. Every restriction Anthropic lobbied for, every self-imposed safety guardrail, every month of delay in releasing American frontier models accelerated this. The Chinese labs were not under those restrictions. They were not going to wait. The composable model future Gavin described, where every enterprise runs a frontier model alongside their own fine-tuned open weight model, is coming regardless of what American labs do next. The question is just whether the open weight half of that stack is American or Chinese. Right now it is Chinese. WATCH THE FULL PODCAST ON The All-In Podcast

Ihtesham Ali

86,044 Aufrufe • vor 21 Tagen