Pinned Tweet
For the next few months (or until I get bored), any experimental film festival accepting submissions is getting an exclusive Stop Making Sense made just for their festival. If you curate a festival and are interested please get in touch.
6
11
63
4,078
Short clip from today's Training Image + Video LoRAs class. I have a few spots available if you'd like to join this week before we get deep into it next Sunday youtu.be/6f3Pru1oWkk?si…
0
0
0
84
everyone's talking about AOC taking Schumers seat, but it might be Mamdani's to lose.
Scoop: Zohran Mamdani has been meeting privately with union leaders, urgently talking up the need for a labor-aligned or leftist candidate to run for president, people familiar with the talks tell me.
He's arguing that progressives can't afford to squander their midterm momentum.
Behind closed doors, Mamdani has also said he's uncertain whether AOC will launch a White House run, the sources said.
axios.com/2026/11/08/sco…
9
3
851
116,235
derrick has started yet another project retweeted
With genAI there is a whole new field of “performative visualizations”. The “artist” understands neither the art nor the subject of what is being visualized, but there is quasi-coherence in the visuals and cultural weight attached to the dataset that it becomes a spectacle.
TLにMarc Ngui氏によるドゥルーズ＝ガタリ『千のプラトー』のダイアグラムが流れてきて、かっK〜と思ったので、試しにシェーダにしてみたら結構イケてるbumblenut.com/drawing/art/pl…
15
33
410
16,164
I've started publicly listing the recordings from my Agentic Creativity course on my YouTube channel youtube.com/watch?v=TWAApJ…
2
3
31
2,098
I’ll be honest. I feel like I never really got to what I think is so interesting about this moment. So we’ll have to try this class again in the new year. But I’ll keep putting up the videos so you can be the judge.
0
0
1
75
derrick has started yet another project retweeted
Each of the 400k particles is a unique, untrained, randomly mutated ~10k param neural net btw 🤗
🔊🦠 Neural-Kimchi-4B 🔊🦠
🔊🦠 Neural-Kimchi-4B 🔊🦠
93
164
1,743
76,888
Cold outreach hiring emails are so funny, they’re always like
“Derrick congratulations on your new film that features sound and image. Really groundbreaking.
Have you ever considered working at an AI startup that puts the elderly in so much debt their great great grandchild commits suicide over it? We’re funded by the Kushners and Satan himself. Let me know!”
“Derrick congratulations on your new film that features sound and image. Really groundbreaking.
Have you ever considered working at an AI startup that puts the elderly in so much debt their great great grandchild commits suicide over it? We’re funded by the Kushners and Satan himself. Let me know!”
0
0
18
642
the politics of AI models are getting weird and fascinating
Introducing Arete-27B: Qwen3.8-27B, post-trained to reinforce Western values.
Chinese open models are currently the strongest available, but they ship with state-aligned behavior in the weights. We built an 803-prompt eval across seven value categories, including freedom of and privacy, and measured how often a model adopts China’s party line.
After fine-tuning on roughly 53 million tokens, Arete-27B's propaganda rate fell from 22% to 1%, in line with the Western models we tested. Its Hermes Bench score shows no significant performance tradeoffs.
The values in an open model are one fine-tune away.
Our eval, weights, and training data are available now along with the full blog post: nousresearch.com/arete
1
1
10
713
No disrespect to them, but hearing designers talk about their work has never made me less excited to try and be a designer again.
3
1
21
2,279