Search

What are you looking for?

60 results for ai

AI·Next-ai-draw-io is a Next.js web application that integrates AI capabilities with draw.io diagrams to allow creation, modification, and enhancement of diagrams via natural language commands and AI-assisted visualization.

GitHub Trending (AI)·github.com··6.5Releasenextjsaidraw-io

AI·This repository contains a curated collection of Codex plugin examples for OpenAI, each including a required plugin.json manifest and optional supporting files for skills, agents, and marketplaces.

GitHub Trending (AI)·github.com··5.2Otheropenaipluginscodex

AI·Apple filed a federal lawsuit on July 10, 2026, accusing OpenAI of trade-secret theft by poaching Apple engineers and exploiting a post-departure access bug; the suit targets OpenAI's hardware ambitions and a possible AI-powered iPhone rival.

36氪 AI·36kr.com··3.0Researchaiappleopenai

AI·OpenAI released GPT-5.6 in three tiers—Sol (flagship), Terra, and Luna—boosting benchmarks in Agent's Last Exam (56.3 score), math, coding, security, and research while slashing token costs and adding Max/Ultra reasoning modes.

36氪 AI·36kr.com··3.0Releaseopenaigptllm

AI·Anthropic's J-space analysis of Claude reveals a small subset of internal representations forming a functional global workspace for flexible reasoning, but the paper explicitly states this has no bearing on AI consciousness and clarifies the core of AI alignment research.

36氪 AI·36kr.com··2.8Opinionaillmanthropic

AI·OpenAI folds Codex into the new ChatGPT desktop app as Work mode, shifting focus to usage-based billing and water-meter-style pricing for AI agents, effectively 'killing' the old standalone Codex product.

36氪 AI·36kr.com··2.8Releaseopenaicodexgpt

AI·Alibaba's Lingbo released six embodied AI models from July 7-10, 2026, introducing 'Embodied Native' — a new paradigm that trains foundation models from scratch on physical-world data, interactions, and causal physics rather than migrating internet LLMs, with LingBot-VA 2.0 as the flagship VLA model achieving real-time 150Hz inference and strong generalization.

36氪 AI·36kr.com··2.7Researchembodied-airoboticsfoundation-model

AI·Reports of Meta selling excess AI compute and Anthropic signing massive data-center deals prove the notion of AI compute oversupply is a myth; actual deliverable frontier capacity remains scarce.

36氪 AI·36kr.com··2.6Researchaicomputedata-center

AI·2025 saw multiple AI companionship app closures including Woebot (150M users, clinical tool shut down due to unsustainable economics) and Dot AI (vision divergence, funds exhausted), with global closures in at least five products and domestic in eight; survivors like Replika (post-data deletion), Character.AI (Google talent acquisition), and China's Xin Bing (ice) lost users amid regulatory scrutiny and high inference costs, revealing unsustainable business models lacking self-reinforcing revenue.

36氪 AI·36kr.com··2.6Researchaicompanionshipmarket

AI·LLM-as-judge scores fluctuate across evaluators even with fixed candidate responses, treating the issue as measurement validity. Across four datasets, scaling Qwen3 from 1.7B to 32B parameters yields only limited adjacent gains, while MiniMax M2-to-M2.7 API upgrades show none; stronger judges reduce but do not eliminate position and verbosity biases, with repeated juries offering little benefit under correlated errors.

arXiv cs.CL·arxiv.org··1.9Researchllm-as-judgeevaluation-reliabilitybias

AI·This position paper reviews LLM-driven formal theorem provers and argues that current systems function mainly as solvers for well-defined problems, not as research agents capable of discovering new theorems or resolving open conjectures at the frontier. It identifies key limitations in datasets, exploration, tools, and collaboration and proposes a roadmap for AI4Math systems to support genuine mathematical research.

arXiv cs.CL·arxiv.org··1.8Researchllmformal-methodsai4math

AI·Hidden Decoding expands each token into n independent streams during continued pretraining of a fixed Transformer backbone, using separate embedding tables and retaining intermediate KV caches for latent computation; Stream-Factorized Attention limits cross-stream mixing to reduce costs. Experiments show frontier 80B and 617B MoE models improve on all benchmarks over matched baselines.

arXiv cs.CL·arxiv.org··1.8Researchllmscalingcontinued-pretraining

AI·LLM-based ASR in regulated domains like banking is limited by privacy and real-speech collection costs; synthetic TTS data is a cost-effective substitute, but acoustic mismatch hinders supervised fine-tuning (SFT). Group Relative Policy Optimization (GRPO) applied solely to synthetic speech reduces WER by 40% relative to SFT (36.71% to 22.09%) and by 45% in the SFT-then-GRPO sequence, by improving behavioral calibration and audio attention rather than representations.

arXiv cs.CL·arxiv.org··1.8Researchspeech-recognitionasrllm

AI·Across five instruction-tuned LLMs, ordinary answer-confidence separates correct from wrong answers on answerable questions but fails to distinguish unanswerable questions (e.g., false-premise ones in CREPE), while a linear probe on hidden states performs the opposite, revealing two distinct abstention axes that remain separable even at 14B scale.

arXiv cs.CL·arxiv.org··1.8Researchllmabstentionllm-safety

AI·DominoTree is a training-free best-first tree draft builder for speculative decoding that scores each candidate node by re-applying Domino's GRU-based causal correction along its specific root-to-node path. It restricts per-node correction to top-M candidates for efficiency and delivers up to 6.6x speedup with the highest mean accept length of any tested method.

arXiv cs.CL·arxiv.org··1.8Researchspeculative-decodingdraftingllm-inference

AI·ICDAR 2026 HIPE-OCRepair competition evaluated LLM-assisted post-correction of noisy OCR from 17th-20th century multilingual (EN/FR/DE) historical newspapers and books. Four teams used zero-shot to fine-tuning approaches; results show significant error reduction but recurring over-correction on low-noise inputs, with a public dataset and evaluation framework released.

arXiv cs.CL·arxiv.org··1.7Researchllmocrpost-correction

AI·XALPHA is a memory-driven AI quant researcher that uses multi-source research memory integrating external financial reports and prior discovery feedback, with Macro Brain for theme planning and archetype selection, Micro Brain for hypothesis-to-code translation and tri-alignment verification, and Cross Brain for feedback consolidation, enabling closed-loop continuous alpha discovery that outperforms baselines on CSI300.

arXiv cs.CL·arxiv.org··1.7Researchllmquantai-research