Posts
Can we make Large Language Models (LLMs) think longer and better? A recent paper explores “test-time scaling” - a fascinating technique that allows us to control an LLM’s reasoning process by manipulating its “thinking time”. This capability was first demonstrated by OpenAI’s o1 model, though without revealing their methodology.
You know those days, right? Either you’re jumping from one zoom call to another with barely time to breathe, or you’re in the zone, fingers flying over the keyboard, crushing a tricky bug. And bam! That’s exactly when you might accidentally leave a Logger statement hanging there as your function’s last word. Oops!
Maybe I’m getting old, maybe that’s my new life in Spain, maybe that’s due to decline of Twitter. But I feel the urge of writing: writing notes, writing thoughts, writing about me and my ideas that don’t survive a week, writing whatever comes to my mind.