Curious about the latest in AI research at @Stanford
? @StanfordAILab (SAIL) is teaming up with @StanfordHAI to continue to build a powerful AI research hub on campus. Prof. Carlos @guestrin was recently announced as the new director of SAIL.
RESEARCH
-
Stanford Unites AI Labs Under New Leadership Direction
By
–
-
Reasoning LLMs: Multiple Implementation Approaches and Research
By
–
Not for now as there are ~100 different ways to implement reasoning. I've written about some of them here: https://
magazine.sebastianraschka.com/p/understandin
g-reasoning-llms
… and plan to follow-up with the latest research in a bit on my blog.
I think Unsloth is a great tool btw! -
Machine Learning and Statistics: The Inference Terminology Confusion
By
–
I'm used to it by now. Back when I was a statistics professor, I always had to start my lectures by clarifying that "inference" in machine learning has a different meaning than in stats…otherwise chaos ensued.
-
Evaluation Crisis: MMLU Obsolete, Need Better AI Metrics
By
–
My reaction is that there is an evaluation crisis. I don't really know what metrics to look at right now. MMLU was a good and useful for a few years but that's long over.
SWE-Bench Verified (real, practical, verified problems) I really like and is great but itself too narrow. -
Terminology Update: Test Set and Test Time Rebranding to Evals and Inference
By
–
test set and test time -> evals and inference
-
AI Code Generation Won’t Match Top Human Programmers by 2026
By
–
> and may match the best human coders by late 2026
No, unless you compare it to a human coder working offline. -

Frances Spence: ENIAC Pioneer and Early Computer Programming
By
–
The late Frances Spence (right) was born #otd in 1922. She was a trailblazer in computer programming as one of the first programmers of ENIAC, the first general-purpose electronic computer.
-
Real-time versus offline content collision in AI systems
By
–
makes sense, esp when you realize real-time and offline content is on a collision course
-

Systematic Survey of Automatic Prompt Optimization Methods
By
–
9). A Systematic Survey of Prompt Optimization This paper offers a comprehensive survey of Automatic Prompt Optimization (APO)—defining its scope, presenting a unifying 5-part framework, categorizing existing methods,…
-

Protein LLMs: Architectures, Training, and Applications Overview
By
–
10). Protein LLMs A comprehensive overview of Protein LLMs, including architectures, training datasets, evaluation metrics, and applications.