Author
Metr
By and about Metr · page 3
Are We in a Bubble? Morgan Stanley Graphic
Morgan Stanley chart compares current valuations to previous market bubbles, showing how today's metrics stack up against the dot-com and 2008 peaks.
newsThe State of AI Safety in 4 Fake Graphs
A researcher uses fabricated graphs to illustrate the absurdity of confident claims about AI safety progress when the underlying metrics don't actually exist.
newsThinking to Recall: How Reasoning Unlocks Parametric Knowledge in LLMs
Google Research finds that enabling reasoning in LLMs helps them recall simple facts through two mechanisms: extra tokens provide computational buffer time, and intermediate facts prime the correct answer—but hallucinated intermediate facts sharply…
newsRevisiting the Platonic Representation Hypothesis: An Aristotelian View
Researchers show the Platonic Representation Hypothesis—the claim that scaled AI models converge on shared internal representations—is largely a statistical artifact of measurement bias, with the apparent convergence disappearing once properly calibrated.
newsFace Recognition in a Few Lines of Python
An engineer demonstrated face recognition in just a few lines of Python using DeepFace and cosine distance to compare facial embeddings across images.
newsNeural networks at scale converge to a shared model of reality
Neural networks trained on different tasks and data converge to shared statistical models of reality, suggesting sufficiently strong learners rediscover natural coordinates in the world rather than inventing arbitrary representations.
newsContinuous Autoregressive Language Models (CALM)
Tencent and Tsinghua propose CALM, a model that predicts continuous vectors representing multiple tokens instead of one token at a time, reducing inference steps by 4× and training compute by 44%.
newsBenchmarks & Metrics: Models Increasingly Overfitted to Leaderboards
Most AI models are heavily overfitted to popular benchmarks, making leaderboard scores a poor proxy for real-world performance and more of a tracker for release velocity.
newsOpenAI Launches GPT-5 with Advanced Reasoning
OpenAI launched GPT-5 on August 7, 2025, claiming PhD-level reasoning and a 70% coding win rate over GPT-4, but the rollout sparked backlash over glitches, hallucinations, and a colder tone, prompting the company to restore legacy models for paid users.
newsBlog Post: The Second Half by Shunyu Yao
OpenAI researcher Shunyu Yao argues that AI's second half will shift from developing new training methods to defining the right problems and evaluation metrics, with pre-training priors mattering more than algorithms.
newsVideo: How AI Image Generation Draws from Physics (Welch Labs)
Welch Labs explains how diffusion models generate images by visualizing the geometry of gradient descent through physics principles.
newsCornell's vec2vec translates embeddings between models it has never seen
Cornell researchers' vec2vec translates text embeddings between models with no paired data or encoder access, hitting up to 0.92 cosine similarity and 100% top-1 accuracy, and recovering readable info from up to 80% of test emails via zero-shot inversion.
newsAnthropic Claude Sonnet 3.7 (Aurora) 🛸
Anthropic released Claude Sonnet 3.7 (Aurora) on March 10, 2025, achieving a ChatEval score of 0.85 and an XGLUE score of 0.88, with the company also introducing prompt caching to reduce inference costs.