That paper is by the same authors – it's about gpt-4o-mini-2024-07-18 though which is a year old now
RESEARCH
-
LLM-as-Judge with Rationale Improves Evaluation Results
By
–
Yeah that makes sense to me – I've been doing some LLM-as-judge thing decently and having it provide the rationale before the winner does appear to provide better results
-

Top AI Researchers Largely Absent from Twitter Social Media
By
–
super cool! inspiring and a great reminder that most of the top AI researchers are not, in fact, on twitter. most do not tweet, many do not have accounts the loudest people on here, the ones who post the most about AI, are typically just random anonymous tech bros
-
Generative Modeling as Subset Classification Framework
By
–
typically that would be considered a subset of generative modeling
-

Chinese Open-Source AI Model Rivals OpenAI and Claude
By
–
Fuckk…this new open-source AI model from China claims to outperform OpenAI’s o3-mini and closely match Claude 4 Opus on their own benchmarks. Huge if true!! SOTA open-source models are raining cats and dogs from China.
-
GPT-5 and the AI models we built together
By
–
Maybe GPT-5 was all the friends we made along the way: Zenith, Summit, Lobster…
-
Research Problems Solved Through New Intellectual Frameworks
By
–
one does not simply “solve” a research problem one instead develops a new intellectual framework from first principles.
asking and answering questions and subquestions, perhaps over several months eventually, upon revisiting the original problem, it is realized to be trivial -

Prompt Hacks Effectiveness: Model Evolution and Measurement Challenges
By
–
The problem with prompting hacks like "I'll tip you $200" is that they mostly emerged a year or two ago and models have evolved a lot since then It's very easy to remain superstitious about them though, especially as measuring their effectiveness is actually very difficult
-
GPT-5 Dominates Lateral Reasoning Benchmark Competition
By
–
The presumably GPT-5 outperforms all competitors in lateral reasoning by a large margin in this new benchmark. The excitement for OpenAI's new model grows daily. https://
x.com/synthwavedd/st
/synthwavedd/status/1951645151203324099
… -
Complete Two-Year AI Degree Curriculum Overview
By
–
hypothetical two-year degree in AI: – Coding in Python
– Semiconductors 101
– Intro to machine learning
– Intro to data science
– Data visualization & dimensionality reduction
– Machine learning engineering
– Language models
– Deep learning (basics & architectures)
–