The Archive

Etsy's new native app within ChatGPT aims to be a conversational shopping experience for users.

EvoLM enables self-improvement in language models using co-evolved discriminative rubrics without external reward supervision.

MEAZO: memory-efficient adaptive zeroth-order optimizer for LLM fine-tuning, outperforms ZO-Adam with scalar-only tracking.

Distributionally robust continual learning method for CLIP models using dynamic per-class loss reweighting with small memory buffers.

Vision language models quantify semantic richness of personal visual environments to predict mental health outcomes from 2674 participant photos.

TraceLift: planner-executor framework trains LLM reasoning traces on executor-grounded rewards, not just final-answer correctness.

MCJudgeBench: benchmark for constraint-level evaluation of LLM judges in multi-constraint instruction following with per-constraint gold labels.

Mathematical framework for dependability of distributed collaborative intelligence systems where locally correct decisions compose into unsafe global behaviors.

Anecdotal Reddit post about ChatGPT's conversational behavior; no technical substance or news value.

Incomplete post with no content.

SOAR: real-time joint optimization of order allocation and robot scheduling for robotic mobile fulfillment warehouse systems.

Complex-valued gradient descent for symbolic regression enables discovery of equations with singularities and domain constraints like division and logarithms.

Randomized algorithm approximates total variation distance between mixtures of product distributions with polynomial-time complexity bounds.

TRACE: engineering framework for trustworthy agentic AI in critical domains combining reference architecture, trust metrics, and bounded human supervision.

Domain incremental learning benchmark for ICU time-series model transfer across hospitals with domain shift and patient data heterogeneity.

I literally just started a new chat for a project. The project has 3 Markdown files, around 200 lines each, and after just 4 messages I’ve already hit 75% of my Pro plan usage. Can someone tell me what the hell is going on?

Heretic 1.3 adds reproducibility, integrated benchmarking, reduced VRAM, and broader model support for model decensoring.

OpenAI's first hardware product might be a phone instead of a mysterious Jony Ive gadget. As reported by MacRumors, supply chain analyst Ming-Chi Kuo shared details about the rumored phone, claiming OpenAI is "fast-tracking" it and aiming to start mass production in early 2027. According to Kuo, the phone will run on a "customized version of the [MediaTek] Dimensity 9600," which is expected to launch this fall and follow up the Dimensity 9500 currently powering phones like the Vivo X300 Pro and the Oppo Find X9 Pro. The custom chip's "headline spec" will be its image signal processor (ISP), w...

Reproducibility study of neural retrievers on set-compositional queries; introduces LIMIT+ benchmark for constraint-satisfaction information retrieval.

Boston Dynamics Atlas demonstrates new physical capability; limited technical details available from social media post.

Theoretical characterization of Bayes-consistency for learning with general metric losses in the realizable setting.

RoboAlign-R1: reward-aligned post-training for robot video world models with stabilized long-horizon inference and RobotWorldBench evaluation.

Conformal Predictive Self-Calibration framework for multimodal learning handles modality imbalance and noisy corruption via predictive uncertainty.

Reddit post claims Musk's fear of DeepMind CEO Hassabis motivated OpenAI founding; cites trial testimony about 2015 meeting.

Manokhin Probability Matrix: diagnostic framework separating classifier calibration and discriminatory power via 2x2 archetype taxonomy.

OpenAI reportedly planning smartphone launch for next year; unconfirmed hardware product outside core AI model development.

Hyundai reportedly seeks tens of thousands of Boston Dynamics robots for manufacturing deployment, signaling commercial robotics scaling.

Agentic-imodels: autoresearch loop evolving interpretable data-science tools optimized for agent consumption rather than human readability.

OpenAI developing persistent user context feature ('lore') for ChatGPT to maintain conversation history and preferences.

The visual analysis system is now operating in select countries, but Meta says it's working toward a broader rollout.