Today Thinking Machines Lab is launching our research blog, Connectionism. Our first blog post is “Defeating Nondeterminism in LLM Inference”
We believe that science is better when shared. Connectionism will cover topics as varied as our research is: from kernel numerics to prompt engineering. Here we share what we are working on and connect with the research community frequently and openly.
The name Connectionism is a throwback to an earlier era of AI; it was the name of the subfield in the 1980s that studied neural networks and their similarity to biological brains.
https://t.co/lrJioBmpbT
Smol tip: pasting long text into Claude Code collapses it into [Pasted text #1]. If you need to edit it you can paste again to expand.
Or, Ctrl+G to open the whole prompt in your code editor. Saved changes sync back to Claude's input field.
another example of how terrible humans are at understanding exponential progress
in 2022, AI experts gave AI a 50% chance of proving publishable math theorems by around 2050 and winning the Putnam math competition by around 2033
LLMs did both this year
🚨SlowMist Security Alert🚨
Recently, victims have been phished by people pretending to be journalists. The scammer often spoke broken Chinese and sends a normal-looking Calendly link. However, upon clicking, the link’s name changes to “Calendly.”, with an additional dot. This link tricks you into authorizing control of your Twitter account, enabling the scammers to post phishing links through your tweets. Similar to incidents encountered by many influencers.
Please be vigilant against unfamiliar links and always stay cautious.
Fable 5 just performed a 6.5x decode performance increase and 7.2x prefill performance increase in the span of a 55 minute goal thus far in my inferencing engine for Nemotron models, using up 8% of my 5-hour limits and 3% of my weekly limits.
And it's found more.
> At @tufalabs, we're specifically interested in multi-turn, long-context interactive environments and *ARC AGI-3* is exactly the type of problem we wanna take on.
> It reduces some of the complexities around safety and setting up environments and focuses on the core problems...so we're super excited
My car isn't getting washed up but @AnthropicAI certainly is 🫵😂
Top is DeepSeek-V4-Flash-DSpark running @ 250TPS on 2 local GPUs
Bottom is Sonnet 5, which tbf should definitely be regulated because it is going to ruin a LOT of codebases