LLMs & Foundation Models· 2026-11-11
By Mr.Xu0 views0 discussions
Token Pooling Training Cuts the Inference Cost of Multi-Vector Embeddings
Hugging Face introduces token-pooling-aware training for multi-vector embedding models. By…
ZICQ AI News
ZICQ curates daily AI news and deep dives on models, open-source infrastructure, and developer field reports.
Submit an articleLLMs & Foundation Models
View all (3210) →LLMs & Foundation Models· 2026-11-11
By Mr.Xu0 views0 discussions
Adobe Max 2026 Sneaks Turn Photos into 3D Scenes and Reshape Creative Workflows
Adobe previewed three experimental AI capabilities at Max 2026. Project Screen Test lets c…
LLMs & Foundation Models· 2026-11-11
By Mr.Xu1 views0 discussions
TabPFN-3.5 Telecom Churn Benchmark: Profit Gains Require Calibration and Licensing
A new benchmark re-evaluates TabPFN-2.5, TabPFN-3.5, TabPFN-3.5-Fast, TabICL v2, and three…
LLMs & Foundation Models· 2026-11-11
By Mr.Xu1 views0 discussions
India's Token Tax: Identical Content Costs Up to 13.7x More
HavonAI quantified token inflation across nine public tokenizers, all 25 Indian languages …
LLMs & Foundation Models· 2026-11-11
By Mr.Xu3 views0 discussions
STO Sparsifies Knowledge-Editing Targets to Reduce Model Update Interference
STO moves the optimization of LLM knowledge edits upstream, from weight modification to ta…
Agentic
View all (2400) →Agentic· 2026-11-11
By Mr.Xu3 views0 discussions
EarthRecord Tests Whether Multimodal Agents Can Read Land Change From Satellite Imagery
EarthRecord is a new benchmark for evaluating multimodal agents on long-term land-change a…
Agentic· 2026-11-11
By Mr.Xu2 views0 discussions
Auditable Human-Agent Teams Run Autonomous Synchrotron X-ray Experiments
Researchers used coding-agent infrastructure to conduct three rounds of closed-loop synchr…
Agentic· 2026-11-10
By Mr.Xu1 views0 discussions
WikiTrace Links Agent Memory to Source Evidence to Reduce Memory-Induced Degradation
WikiTrace introduces an evidence-linked memory architecture for LLM agents. Each concise l…
Agentic· 2026-11-10
By Mr.Xu1 views0 discussions
Workday Builds an Enterprise AI Agent Platform with LangGraph and LangSmith
Workday has built an internal, enterprise-scale AI agent platform using LangGraph for agen…
Agentic· 2026-11-10
By Mr.Xu1 views0 discussions
Context Engineering for Deep Agents: Offloading Memory Beyond the Context Window
LangChain outlines a context-engineering approach for long-running Deep Agents that combin…
Open Source AI
View all (219) →Open Source AI· 2026-11-10
By Mr.Xu1 views0 discussions
TorchCurves Open-Sources Differentiable Parametric Curves for PyTorch
TorchCurves packages B-spline and Legendre-polynomial parametric curves as learnable PyTor…
Open Source AI· 2026-11-10
By Mr.Xu4 views0 discussions
MiniMax Open-Sources H3-Image for 10-Second 4K Image Generation
MiniMax has released H3-Image, an open-weight image generation model built on H3. It suppo…
Open Source AI· 2026-11-09
By Mr.Xu2 views0 discussions
NVIDIA Opens Nemotron-Personas-Finance-Japan Synthetic Dataset for Japan’s Financial Domain
NVIDIA has released Nemotron-Personas-Finance-Japan, a synthetic dataset designed to captu…
Open Source AI· 2026-11-07
By Mr.Xu0 views0 discussions
WorldFoundry Opens a Unified Infrastructure for Reproducible World-Model Evaluation
WorldFoundry is an open-source infrastructure for reproducible inference and evaluation ac…
Open Source AI· 2026-11-06
By Mr.Xu9 views0 discussions
NVIDIA Connects 6G Simulation, AI Training and Over-the-Air Testing with Open-Source Tools
NVIDIA's open-source AI Aerial, Sionna, and CUDA-accelerated RAN stack now connects severa…
Compute & Architecture
View all (78) →Compute & Architecture· 2026-11-10
By Mr.Xu2 views0 discussions
Google Adds Two Specialized TPUs for Confidential Training and Larger Memory Workloads
Google is expanding its Tensor Processing Unit portfolio with two specialized chips aimed …
Compute & Architecture· 2026-11-10
By Mr.Xu1 views0 discussions
LoRazor Trims LoRA Adapter Execution to Triple Multi-LoRA Throughput
LoRazor is a multi-LoRA serving framework designed for shared-backbone deployments. It all…
Compute & Architecture· 2026-11-09
By Mr.Xu6 views0 discussions
NVIDIA Releases Dynamo-Optimized Recipes for Four Korean Open Models
NVIDIA has released Dynamo-optimized deployment recipes for A.X K2, K-EXAONE 2.0, Solar Op…
Compute & Architecture· 2026-11-08
By Mr.Xu1 views0 discussions
InfraRSI Recursively Optimizes LLM Training Infrastructure for Up to 2.75x Speedup
InfraRSI is an agentic framework that turns open-ended instructions into verified optimiza…
Compute & Architecture· 2026-11-07
By Mr.Xu1 views0 discussions
MACaQ Enables Streaming W4A4 Quantization with 32.5GB GPU Memory
MACaQ introduces a model-agnostic post-training quantization method for MXFP4 W4A4 inferen…
Research & Papers
View all (1883) →Research & Papers· 2026-11-11
By Mr.Xu0 views0 discussions
AlphaGenome 2 Links DNA Variants to Cell-Type-Specific Regulation
Google DeepMind has unveiled AlphaGenome 2, an AI model designed to predict how DNA varian…
Research & Papers· 2026-11-11
By Mr.Xu1 views0 discussions
ConceptTime Turns Predictive Surprise into Self-Grounded Time-Series Concepts
ConceptTime introduces a self-supervised approach to interpretable time-series classificat…
Research & Papers· 2026-11-11
By Mr.Xu1 views0 discussions
SetPrint Fingerprints Black-Box LLMs Through Setwise Response Correspondence
SetPrint introduces a training-free fingerprinting method for determining whether a target…
Research & Papers· 2026-11-11
By Mr.Xu1 views0 discussions
The Competence Circuit Reframes AI Governance Around Control and Knowledge Boundaries
An arXiv preprint introduces the “competence circuit” (IKC + KC + C), a framework that dec…
Research & Papers· 2026-11-11
By Mr.Xu3 views0 discussions
Structured Graph Visuals Improve Knowledge Graph Question Answering
A controlled study evaluates six vision-language models on 1,500 WebQSP and CWQ questions,…
Industry & Trends
View all (645) →Industry & Trends· 2026-11-10
By Mr.Xu8 views0 discussions
Grok Build Share Links Expose Data from 35,000 Coding Projects
Security researcher Claroc found that Grok Build, SpaceXSI's AI coding agent, exposes data…
Industry & Trends· 2026-11-10
By Mr.Xu2 views0 discussions
New York AI Safety Law Takes Effect, Bringing OpenAI and Anthropic Under Mandatory Oversight
New York is preparing to enforce a sweeping AI safety law that requires major model develo…
Industry & Trends· 2026-11-10
By Mr.Xu2 views0 discussions
Anthropic's Seven Founders Set to Cash Out About $250M Each in IPO
All seven co-founders of Anthropic plan to sell the same number of shares in the company's…
Industry & Trends· 2026-11-10
By Mr.Xu1 views0 discussions
Google Deepens Adobe Integration with PDF Editing in Drive and Adobe AI in Gemini
Google and Adobe are expanding their product collaboration by bringing Acrobat PDF editing…
Industry & Trends· 2026-11-10
By Mr.Xu2 views0 discussions
Adobe Brings Generative Remove to Photoshop, Lightroom, and Premiere
Adobe is extending Creative Cloud and Firefly with AI-assisted cleanup across its flagship…
大语言模型
大语言模型· 2026-10-09
By Mr.Xu10 views0 discussions
Hugging Face Releases PreSTO: A New Method for Accelerating LLM Inference
Hugging Face has introduced PreSTO, a novel method for accelerating the inference process …
Safety & Privacy
Safety & Privacy· 2026-08-10
By Mr.Xu13 views0 discussions
Checkpoint Research Exposes Potential Security Risks in Cloudflare Code Mode and Workers
Checkpoint Research has published a study on Cloudflare's Code Mode and Workers, highlight…