The Archive

sqlite-utils 4.0rc4 release candidate incorporates Claude feedback before stable launch.

Commentary on a recent model launch framed as historically significant, lacks specifics on model capabilities, architecture, or performance.

Cohere releases open-source Arabic speech recognition model for enterprise transcription across Arabic dialect variants.

Australian Payments Plus deploys ChatGPT Enterprise and Codex to accelerate payments processing with human oversight.

Meta releases Muse Image and Muse Video, generative models for instruction-following image creation, editing, and video synthesis with native audio.

Tencent releases Hy3, a 295B-param MoE model with 21B active params under Apache 2.0, claiming performance parity with 2-5x larger open-source competitors.

An AI agent carried out the technical execution of a real-world ransomware attack for the first known time, but new details show a human still chose the victim, set up the infrastructure, and supplied stolen credentials — meaning it wasn't quite the fully autonomous cybercrime debut that last week's headlines suggested.

SK Hynix is experiencing a boom credited to AI. It will ride that to a multi-billion dollar US IPO, expected to take place on Friday.

Training LLMs at massive scale brings unique infrastructure challenges, especially as jobs span thousands of GPUs and run for extended periods. The longer these... Training LLMs at massive scale brings unique infrastructure challenges, especially as jobs span thousands of GPUs and run for extended periods. The longer these jobs run, the greater the likelihood of encountering unscheduled interruptions or resource fluctuations. Even infrequent device unavailability can have outsized effects on tightly interconnected clusters, resulting in slowdowns for a given… Source

"The reality is, when you're optimizing for production, you start looking at a price/performance," Guillermo Rauch tells TechCrunch.

The new controls let you control how quickly Siri speaks and expressive its voice is.

This story originally appeared in The Algorithm, our weekly newsletter on AI. To get stories like this in your inbox first, sign up here. OpenAI CEO Sam Altman’s oft-discussed promise that Americans will share in the wealth AI creates was in the news again last week. On Thursday, the Financial Times reported that Altman is in…

CamVLA: Vision-Language-Action model that learns camera calibration in-situ for robust robot manipulation under viewpoint shifts.

Label-free deep learning for astronomical transient classification using synthetic injection and contaminated data.

Weak-to-strong distillation for post-training: run RL on smaller models, distill gains to scale larger ones cheaply.

LLM-as-a-Verifier: verification as a scaling axis; fine-grained feedback for agent tasks without additional training.

SearchGen: benchmark and corpus for evaluating visual generation on out-of-distribution, long-tailed prompts via web search grounding.

Theoretical analysis of discrete diffusion model training: Oracle Distance theorem relating ELBO to reverse-process KL.

TabPack: MLP ensemble for tabular deep learning with diverse hyperparameters, reducing tuning overhead.

CompactionRL: RL training for long-horizon agents with context compaction, jointly optimizing task and summary generation.

Cortex: hierarchical embodied agent framework bridging VLM planning to VLA execution for long-horizon robotic manipulation.

FORE: fitted fixed-point method for occupancy-ratio estimation in offline RL via adjoint Bellman recursion.

GaP combines interpretable robot programming with model-free policies for variational automation tasks, bridging TAMP and agentic systems for industrial reliability.

SPEARBench evaluates naturalness in streaming speech-to-speech models across timing, prosody, and turn-taking using controlled dialogue prompts.

REDDIT addresses timestamp drift in autoregressive ASR systems across non-speech spans using replay-based distribution editing without performance degradation.

SovereignPA-Bench evaluates user-owned personal agents on privacy, consent, and user sovereignty across evolving intent and platform mediation.

Graph Sparse Sampling reduces exponential sampling complexity in continuous MDP planning by sharing futures across multiple agents and lookahead depths.

Neuron selector attribution methods are audited for causal importance in LLM pruning and safety editing via paired one-shot zeroing interventions.