Pete Werner
pete.penumbra.software
Pete Werner
@pete.penumbra.software
Founder at Penumbra AI
Previously Head of AI at Leonardo, acquired by Canva 2024.
MSc Mathematical and Statistical modeling. AWS Certified Architect.
AI • Art • Music • Yoga • Cycling
Am I missing something here or did they train a model to spout gibberish after a specific rare token then consider it noteworthy when it works? www.anthropic.com/research/sma...
A small number of samples can poison LLMs of any size
Anthropic research on data-poisoning attacks in large language models
www.anthropic.com
October 11, 2025 at 11:46 PM
The impressive thing about Gen AI is how often it actually works
October 1, 2025 at 11:36 PM
Great write up on matmuls if you’re into the gory details www.aleksagordic.com/blog/matmul
Inside NVIDIA GPUs: Anatomy of high performance matmul kernels - Aleksa Gordić
From GPU architecture and PTX/SASS to warp-tiling and deep asynchronous tensor core pipelines.
www.aleksagordic.com
October 1, 2025 at 11:16 PM
I will be visiting Atlassian in a few weeks for a panel discussion on Reinforcement Learning, come along if you’re in Sydney www.aicamp.ai/event/eventd...
AI Meetup (Sydney) with Atlassian - Reinforced Learning for AI Models
Join over half million developers learning how to use and build AI through expert-led tech talks, workshops, bootcamps and crash courses. Level up your skills, and stay ahead of the industry | AICamp
www.aicamp.ai
September 29, 2025 at 2:24 AM
Another banger from Eugene Yan
I've been nerdsniped by the idea of Semantic IDs.

Here's the result of my training runs:
• RQ-VAE to compress item embeddings into tokens
• SASRec to predict the next item (i.e., 4-tokens) exactly
• Qwen3-8B that can return recs and natural language!

eugeneyan.com/writing/sema...
How to Train an LLM-RecSys Hybrid for Steerable Recs with Semantic IDs
An LLM that can converse in English & item IDs, and make recommendations w/o retrieval or tools.
eugeneyan.com
September 29, 2025 at 2:22 AM
Feel like I don’t hear AGI as much as I did 3-6 months ago. I guess the checks have cleared.
September 24, 2025 at 11:15 PM
Fleshing out a proposal with ChatGPT: 5 minutes
Validating the details: 4 hours
September 13, 2025 at 3:17 AM
If you feel old, ChatGPT just told me “you’re among the ancient ones of the web.”
August 2, 2025 at 10:26 AM
Startup idea: Secure MCP. It’s just mcp but the logo is a padlock.
June 23, 2025 at 1:39 AM
Hot take: Apple is second only to NVIDIA when it comes to AI. They have been doing it a long time, their own hardware and importantly mature and robust software on top of it. #wwdc
June 9, 2025 at 7:25 AM
I aspire to the level of brazenness whoever makes the marketing charts for NVIDIA has attained
June 4, 2025 at 3:48 AM
Remember in 2016 people were going to hail a self driving Uber instead of owning a car and driving themselves
May 23, 2025 at 6:54 AM
RIP Civit
May 22, 2025 at 9:52 PM
An ablation study is not mathematical rigor. It’s an empirical experiment.
May 22, 2025 at 11:22 AM
If you can’t think of any good use cases for LLMs maybe you’re just boring and uncreative
May 7, 2025 at 1:47 AM
If you are in Sydney this April 30 I will be giving a talk on scaling up AI services at AI Camp in Sydney. How we built and scaled the core AI services that drove our product to over 10 million users. Be sure to come along if it sounds of interest. www.aicamp.ai/event/eventd...
AI Meetup (Sydney): GenAI, LLMs and Agent
Join over half million developers learning how to use and build AI through expert-led tech talks, workshops, bootcamps and crash courses. Level up your skills, and stay ahead of the industry | AICamp
www.aicamp.ai
April 23, 2025 at 5:13 AM
Fantastic run through of the core pointy end of flow matching youtu.be/7cMzfkWFWhI
Flow Matching | Explanation + PyTorch Implementation
YouTube video by Outlier
youtu.be
April 21, 2025 at 2:38 AM
First post 😎
April 19, 2025 at 12:41 AM