@ericschwarzkopf.bsky.social
Reposted
A short talk on the main architecture components of LLMs this year + a look beyond the transformer architecture: www.youtube.com/watch?v=lONy...
October 27, 2025 at 3:45 PM
A short talk on the main architecture components of LLMs this year + a look beyond the transformer architecture: www.youtube.com/watch?v=lONy...
Reposted
From the Hierarchical Reasoning Model (HRM) to a new Tiny Recursive Model (TRM).
A few months ago, the HRM made big waves in the AI research community as it showed really good performance on the ARC challenge despite its small 27M size. (That's about 22x smaller than the smallest Qwen3 0.6B model.)
A few months ago, the HRM made big waves in the AI research community as it showed really good performance on the ARC challenge despite its small 27M size. (That's about 22x smaller than the smallest Qwen3 0.6B model.)
October 9, 2025 at 4:23 PM
From the Hierarchical Reasoning Model (HRM) to a new Tiny Recursive Model (TRM).
A few months ago, the HRM made big waves in the AI research community as it showed really good performance on the ARC challenge despite its small 27M size. (That's about 22x smaller than the smallest Qwen3 0.6B model.)
A few months ago, the HRM made big waves in the AI research community as it showed really good performance on the ARC challenge despite its small 27M size. (That's about 22x smaller than the smallest Qwen3 0.6B model.)
Reposted
How do we evaluate LLMs?
I wrote up a new article on
(1) multiple-choice benchmarks,
(2) verifiers,
(3) leaderboards, and
(4) LLM judges
All with from-scratch code examples, of course!
sebastianraschka.com/blog/2025/ll...
I wrote up a new article on
(1) multiple-choice benchmarks,
(2) verifiers,
(3) leaderboards, and
(4) LLM judges
All with from-scratch code examples, of course!
sebastianraschka.com/blog/2025/ll...
Understanding the 4 Main Approaches to LLM Evaluation (From Scratch)
Multiple-Choice Benchmarks, Verifiers, Leaderboards, and LLM Judges with Code Examples
sebastianraschka.com
October 5, 2025 at 3:51 PM
How do we evaluate LLMs?
I wrote up a new article on
(1) multiple-choice benchmarks,
(2) verifiers,
(3) leaderboards, and
(4) LLM judges
All with from-scratch code examples, of course!
sebastianraschka.com/blog/2025/ll...
I wrote up a new article on
(1) multiple-choice benchmarks,
(2) verifiers,
(3) leaderboards, and
(4) LLM judges
All with from-scratch code examples, of course!
sebastianraschka.com/blog/2025/ll...
Reposted
40 years ago; C64 Paradroid was completed and sent off for duplication. I`m still amazed both at how quickly it was put together and how it has stood the test of time. I`ll be celebrating Paradroid until Christmas at least.
September 8, 2025 at 12:00 PM
40 years ago; C64 Paradroid was completed and sent off for duplication. I`m still amazed both at how quickly it was put together and how it has stood the test of time. I`ll be celebrating Paradroid until Christmas at least.
Reposted
Here we go again! 🤩 🚀 💿
coldstorage.bandcamp.com/album/wipeou...
coldstorage.bandcamp.com/album/wipeou...
wipE'out'' - The Zero Gravity Soundtrack Vol. 2, by CoLD SToRAGE
22 track album
coldstorage.bandcamp.com
September 7, 2025 at 10:24 PM
Here we go again! 🤩 🚀 💿
coldstorage.bandcamp.com/album/wipeou...
coldstorage.bandcamp.com/album/wipeou...
Reposted
I had a great week at Gamescom, but critically I was seeing and hearing more meaningful discussion of where AI sits in the industry beyond hype and outright rejection.
Scribbled down some thoughts on this in this week's newsletter.
Scribbled down some thoughts on this in this week's newsletter.
Newsletter update - Catch up on all that is happening in the games industry and AI. This week's special is a closer look at Devcom and Gamescom which @tommy.aiandgames.com attended last week.
www.aiandgames.com/p/reflection...
#gamescom #devcom #gamedev #indiegames #AIandGames
www.aiandgames.com/p/reflection...
#gamescom #devcom #gamedev #indiegames #AIandGames
Reflections on Gamescom 2025
I keep hearing the term 'AI' getting thrown around...
www.aiandgames.com
August 28, 2025 at 6:08 PM
I had a great week at Gamescom, but critically I was seeing and hearing more meaningful discussion of where AI sits in the industry beyond hype and outright rejection.
Scribbled down some thoughts on this in this week's newsletter.
Scribbled down some thoughts on this in this week's newsletter.
Reposted
Headline of the Day:
August 6, 2025 at 7:14 PM
Headline of the Day:
Reposted
Motherfucking wind farms…
July 30, 2025 at 5:02 PM
Motherfucking wind farms…
Reposted
For that weird, narrow segment of technically-minded programmers who want to know how we programmed 3D graphics in the 90s and also want to know about 80% of the stuff I programmed before RAD, I sat down for a 5 hour interview on the Wookash Podcast.
www.youtube.com/watch?v=T1tX...
www.youtube.com/watch?v=T1tX...
Legendary Software Rendering Era | Sean Barrett
YouTube video by Wookash Podcast
www.youtube.com
July 19, 2025 at 4:27 PM
For that weird, narrow segment of technically-minded programmers who want to know how we programmed 3D graphics in the 90s and also want to know about 80% of the stuff I programmed before RAD, I sat down for a 5 hour interview on the Wookash Podcast.
www.youtube.com/watch?v=T1tX...
www.youtube.com/watch?v=T1tX...
Reposted
Our economic/political systems produce utterly inept leaders. Paid off for failing.
* world is on fire
* authoritarian surveillance state
* medical system sucks
* a tiny percent own everything
* captured media
* shit education
* laughable retirement
* US being dismantled
* tax cuts for the rich
* world is on fire
* authoritarian surveillance state
* medical system sucks
* a tiny percent own everything
* captured media
* shit education
* laughable retirement
* US being dismantled
* tax cuts for the rich
MEDITERRANEAN HEAT WAVE
Another day with up to 44C in North Africa and 36C in the Balkans.
Some records for early June fell including Debrecen in Hungary with 34.6C (earliest 35C was on a 20 June).
Tomorrow extreme heat in North Africa,Spain and the Balkans
Another day with up to 44C in North Africa and 36C in the Balkans.
Some records for early June fell including Debrecen in Hungary with 34.6C (earliest 35C was on a 20 June).
Tomorrow extreme heat in North Africa,Spain and the Balkans
June 7, 2025 at 11:06 PM
Our economic/political systems produce utterly inept leaders. Paid off for failing.
* world is on fire
* authoritarian surveillance state
* medical system sucks
* a tiny percent own everything
* captured media
* shit education
* laughable retirement
* US being dismantled
* tax cuts for the rich
* world is on fire
* authoritarian surveillance state
* medical system sucks
* a tiny percent own everything
* captured media
* shit education
* laughable retirement
* US being dismantled
* tax cuts for the rich
Reposted
**NEW VIDEO**
We celebrate 80(!?!?) episodes of @aiandgames.com by exploring how Massive Entertainment trained neural networks to control the NPC speeders in 2024's Star Wars Outlaws.
Including a ton of behind-the-scenes footage of the AI training process
Watch on YouTube
Read on our Website
We celebrate 80(!?!?) episodes of @aiandgames.com by exploring how Massive Entertainment trained neural networks to control the NPC speeders in 2024's Star Wars Outlaws.
Including a ton of behind-the-scenes footage of the AI training process
Watch on YouTube
Read on our Website
Building AI Speeders with Machine Learning in Star Wars Outlaws | AI and Games #80
YouTube video by AI and Games
www.youtube.com
May 30, 2025 at 1:03 PM
**NEW VIDEO**
We celebrate 80(!?!?) episodes of @aiandgames.com by exploring how Massive Entertainment trained neural networks to control the NPC speeders in 2024's Star Wars Outlaws.
Including a ton of behind-the-scenes footage of the AI training process
Watch on YouTube
Read on our Website
We celebrate 80(!?!?) episodes of @aiandgames.com by exploring how Massive Entertainment trained neural networks to control the NPC speeders in 2024's Star Wars Outlaws.
Including a ton of behind-the-scenes footage of the AI training process
Watch on YouTube
Read on our Website
Reposted
Just shared a new article on "The State of Reinforcement Learning for LLM Reasoning"!
If you are new to reinforcement learning, this article has a generous intro section (PPO, GRPO, etc)
Also, I cover 15 recent articles focused on RL & Reasoning.
🔗 magazine.sebastianraschka.com/p/the-state-...
If you are new to reinforcement learning, this article has a generous intro section (PPO, GRPO, etc)
Also, I cover 15 recent articles focused on RL & Reasoning.
🔗 magazine.sebastianraschka.com/p/the-state-...
April 19, 2025 at 1:48 PM
Just shared a new article on "The State of Reinforcement Learning for LLM Reasoning"!
If you are new to reinforcement learning, this article has a generous intro section (PPO, GRPO, etc)
Also, I cover 15 recent articles focused on RL & Reasoning.
🔗 magazine.sebastianraschka.com/p/the-state-...
If you are new to reinforcement learning, this article has a generous intro section (PPO, GRPO, etc)
Also, I cover 15 recent articles focused on RL & Reasoning.
🔗 magazine.sebastianraschka.com/p/the-state-...
Reposted
The Spring issue of 2600 has been released! www.2600.com/content/spri...
SPRING ISSUE OF 2600 RELEASED | 2600
www.2600.com
April 16, 2025 at 7:51 PM
The Spring issue of 2600 has been released! www.2600.com/content/spri...
Reposted
Coded Llama 3.2 model from scratch and shared it on the HF Hub.
Why? Because I think 1B & 3B models are great for experimentation, and I wanted to share a clean, readable implementation for learning and research: huggingface.co/rasbt/llama-...
Why? Because I think 1B & 3B models are great for experimentation, and I wanted to share a clean, readable implementation for learning and research: huggingface.co/rasbt/llama-...
March 31, 2025 at 5:13 PM
Coded Llama 3.2 model from scratch and shared it on the HF Hub.
Why? Because I think 1B & 3B models are great for experimentation, and I wanted to share a clean, readable implementation for learning and research: huggingface.co/rasbt/llama-...
Why? Because I think 1B & 3B models are great for experimentation, and I wanted to share a clean, readable implementation for learning and research: huggingface.co/rasbt/llama-...
Reposted
My next tutorial on pretraining an LLM from scratch is now out. It starts with a step-by-step walkthrough of understanding, calculating, and optimizing the loss. After training, we update the text generation function with temperature scaling and top-k sampling: www.youtube.com/watch?v=Zar2...
March 23, 2025 at 1:38 PM
My next tutorial on pretraining an LLM from scratch is now out. It starts with a step-by-step walkthrough of understanding, calculating, and optimizing the loss. After training, we update the text generation function with temperature scaling and top-k sampling: www.youtube.com/watch?v=Zar2...
Reposted
🚨 SALE + NEW CONTENT! 🚨
🔥 30% off all courses! 🔥
Just dropped a massive GPGPU update in the Shader Course:
⚡️ Almost 6 hours of new content
⚡️ Basic -> Advanced particle systems
⚡️ Building a 1 MILLION+ boid simulation
👉 Grab it this weekend!
Use this coupon: PARTICLES2025
simondev.io/courses
🔥 30% off all courses! 🔥
Just dropped a massive GPGPU update in the Shader Course:
⚡️ Almost 6 hours of new content
⚡️ Basic -> Advanced particle systems
⚡️ Building a 1 MILLION+ boid simulation
👉 Grab it this weekend!
Use this coupon: PARTICLES2025
simondev.io/courses
March 14, 2025 at 7:33 PM
🚨 SALE + NEW CONTENT! 🚨
🔥 30% off all courses! 🔥
Just dropped a massive GPGPU update in the Shader Course:
⚡️ Almost 6 hours of new content
⚡️ Basic -> Advanced particle systems
⚡️ Building a 1 MILLION+ boid simulation
👉 Grab it this weekend!
Use this coupon: PARTICLES2025
simondev.io/courses
🔥 30% off all courses! 🔥
Just dropped a massive GPGPU update in the Shader Course:
⚡️ Almost 6 hours of new content
⚡️ Basic -> Advanced particle systems
⚡️ Building a 1 MILLION+ boid simulation
👉 Grab it this weekend!
Use this coupon: PARTICLES2025
simondev.io/courses
Reposted
Just read Gemma 3 is out. Gemma models are super underestimated imho. Will be taking this for a spin in the next few days. In the meantime, they have a technical report here: storage.googleapis.com/deepmind-med...
storage.googleapis.com
March 12, 2025 at 4:06 PM
Just read Gemma 3 is out. Gemma models are super underestimated imho. Will be taking this for a spin in the next few days. In the meantime, they have a technical report here: storage.googleapis.com/deepmind-med...
Reposted
I just shared a new article, "The State of Reasoning Models", where I am exploring 12 new research articles on improving the reasoning capabilities of LLMs (all published after the release of DeepSeek R1): magazine.sebastianraschka.com/p/state-of-l...
Happy reading!
Happy reading!
The State of LLM Reasoning Models
Part 1: Inference-Time Compute Scaling Methods
magazine.sebastianraschka.com
March 8, 2025 at 2:37 PM
I just shared a new article, "The State of Reasoning Models", where I am exploring 12 new research articles on improving the reasoning capabilities of LLMs (all published after the release of DeepSeek R1): magazine.sebastianraschka.com/p/state-of-l...
Happy reading!
Happy reading!
Reposted
Correct. If the birthdate field contains corrupt or mismatched data, it defaults to 1875-05-20, which serves as a flag. May 20, 1875, is the day the international standards and metrics treaty was signed. Everything is a conspiracy when you don’t know how anything works.
February 15, 2025 at 5:53 AM
Correct. If the birthdate field contains corrupt or mismatched data, it defaults to 1875-05-20, which serves as a flag. May 20, 1875, is the day the international standards and metrics treaty was signed. Everything is a conspiracy when you don’t know how anything works.
Reposted
AI and Games is back on YouTube!
We dig into the design philosophy of Saber Interactive's 𝐖𝐚𝐫𝐡𝐚𝐦𝐦𝐞𝐫 𝟒𝟎𝐤 𝐒𝐩𝐚𝐜𝐞 𝐌𝐚𝐫𝐢𝐧𝐞 𝟐, and how the game's AI is built both to reward aggression, and punish heresy!
Watch on YouTube
www.youtube.com/watch?v=3g8E...
Read on Substack
www.aiandgames.com/p/how-warham...
We dig into the design philosophy of Saber Interactive's 𝐖𝐚𝐫𝐡𝐚𝐦𝐦𝐞𝐫 𝟒𝟎𝐤 𝐒𝐩𝐚𝐜𝐞 𝐌𝐚𝐫𝐢𝐧𝐞 𝟐, and how the game's AI is built both to reward aggression, and punish heresy!
Watch on YouTube
www.youtube.com/watch?v=3g8E...
Read on Substack
www.aiandgames.com/p/how-warham...
January 30, 2025 at 11:03 AM
AI and Games is back on YouTube!
We dig into the design philosophy of Saber Interactive's 𝐖𝐚𝐫𝐡𝐚𝐦𝐦𝐞𝐫 𝟒𝟎𝐤 𝐒𝐩𝐚𝐜𝐞 𝐌𝐚𝐫𝐢𝐧𝐞 𝟐, and how the game's AI is built both to reward aggression, and punish heresy!
Watch on YouTube
www.youtube.com/watch?v=3g8E...
Read on Substack
www.aiandgames.com/p/how-warham...
We dig into the design philosophy of Saber Interactive's 𝐖𝐚𝐫𝐡𝐚𝐦𝐦𝐞𝐫 𝟒𝟎𝐤 𝐒𝐩𝐚𝐜𝐞 𝐌𝐚𝐫𝐢𝐧𝐞 𝟐, and how the game's AI is built both to reward aggression, and punish heresy!
Watch on YouTube
www.youtube.com/watch?v=3g8E...
Read on Substack
www.aiandgames.com/p/how-warham...
Reposted
Powering the world with clean energy is common sense. It will secure the future for our grandchildren. The clean energy transformation is - and remains - unstoppable www.globalcitizen.org/en/content/t...
Trump Withdraws from Paris Agreement (Again): What This Moment Means for Climate Action
Trump has pulled the US out of the Paris Agreement (again) — but here’s how the US can still fight the climate crisis.
www.globalcitizen.org
January 22, 2025 at 12:54 PM
Powering the world with clean energy is common sense. It will secure the future for our grandchildren. The clean energy transformation is - and remains - unstoppable www.globalcitizen.org/en/content/t...
Reposted
Just launched a Humble Bundle. www.humblebundle.com/books/hackin...
Humble Tech Book Bundle: Hacking 2024 by No Starch
Level up your hacking and skills with this tech bundle from No Starch. Learn to protect yourself and others! Pay what you want & support charity!
www.humblebundle.com
December 2, 2024 at 7:07 PM
Just launched a Humble Bundle. www.humblebundle.com/books/hackin...
Reposted
I do believe that when the bullshit dies down the 'good' stuff will become apparent, but good lord the techbros are working hard to prevent that from happening.
I think my last newsletter of the year really needs to be about the toxic AI image problem and the lasting impact it will have.
I think my last newsletter of the year really needs to be about the toxic AI image problem and the lasting impact it will have.
November 30, 2024 at 11:19 PM
I do believe that when the bullshit dies down the 'good' stuff will become apparent, but good lord the techbros are working hard to prevent that from happening.
I think my last newsletter of the year really needs to be about the toxic AI image problem and the lasting impact it will have.
I think my last newsletter of the year really needs to be about the toxic AI image problem and the lasting impact it will have.
Reposted
Outlined an AI research review article for December… I love traveling but I also can’t wait to be back on my computer 😅.
In the meantime, if you are curious how Multimodal LLMs work, I recently wrote an article to explain the main & recent approaches: magazine.sebastianraschka.com/p/understand...
In the meantime, if you are curious how Multimodal LLMs work, I recently wrote an article to explain the main & recent approaches: magazine.sebastianraschka.com/p/understand...
Understanding Multimodal LLMs
An introduction to the main techniques and latest models
magazine.sebastianraschka.com
November 24, 2024 at 6:54 AM
Outlined an AI research review article for December… I love traveling but I also can’t wait to be back on my computer 😅.
In the meantime, if you are curious how Multimodal LLMs work, I recently wrote an article to explain the main & recent approaches: magazine.sebastianraschka.com/p/understand...
In the meantime, if you are curious how Multimodal LLMs work, I recently wrote an article to explain the main & recent approaches: magazine.sebastianraschka.com/p/understand...