ZICQ AI News

LLMs & Foundation Models· 2026-10-02

arXiv Introduces FedCausalCompose Framework to Enhance Causal Reasoning in Modular LLM Agents

arXiv has released a study on causal world models for modular LLM agents, introducing the …

Mr.Xu0 views0 discussions

LLMs & Foundation Models· 2026-10-02

arXiv Releases Holistic Carbon Footprint Assessment for Noor, a Large Arabic Language Model

ArXiv has released a study assessing the holistic carbon footprint of Noor, an ongoing pro…

Mr.Xu0 views0 discussions

LLMs & Foundation Models· 2026-10-02

arXiv Releases KoDialectBench: A New Benchmark for Assessing Synthetic Korean Dialect Data Recovery

arXiv has released a new study on the recovery effectiveness of synthetic Korean dialect d…

Mr.Xu0 views0 discussions

LLMs & Foundation Models· 2026-10-02

arXiv Releases FACET: A Novel Reference-Free Approach for Machine Translation Quality Evaluation

arXiv has introduced FACET, a novel reference-free approach for machine translation qualit…

Mr.Xu0 views0 discussions

LLMs & Foundation Models· 2026-10-02

arXiv Releases BudgetSchemaBench: A New Diagnostic Tool for Schema Context in Text-to-SQL

arXiv has released a new study on handling database schema context in text-to-SQL tasks, i…

Mr.Xu0 views0 discussions

LLMs & Foundation Models· 2026-10-02

arXiv Research: KLUE-BERT Achieves 99.3% Accuracy in Korean Sexual Offense Legal Text Classification

This study evaluates legal text classification models ranging from traditional machine lea…

Mr.Xu0 views0 discussions

LLMs & Foundation Models· 2026-10-02

arXiv Publishes Study on On-Device NER: Evaluating Accuracy, Cost, Reliability, and Confidence

arXiv has released a study on on-device Named-Entity Recognition (NER), evaluating nine sy…

Mr.Xu0 views0 discussions

LLMs & Foundation Models· 2026-10-01

Where-OPD Released: Spatial Guidance Enables Breakthrough in MLLM Self-Distillation for Enhanced Perception

Where-OPD introduces a novel on-policy self-distillation approach for Multimodal Large Lan…

Mr.Xu0 views0 discussions

LLMs & Foundation Models· 2026-10-01

Hugging Face Releases LoopCD: Revolutionizing Decoding Efficiency for Looped Transformers

Hugging Face has introduced LoopCD, an innovative contrastive decoding framework designed …

Mr.Xu0 views0 discussions

LLMs & Foundation Models· 2026-10-01

Hugging Face Introduces SAGO Framework: Multi-Dimensional Evaluation of LLM Generalization Stability

Hugging Face's research team introduces the Stability-Aware Generalization Objective (SAGO…

Mr.Xu0 views0 discussions

LLMs & Foundation Models· 2026-10-01

Latent-Foresight Framework Released: Enabling End-to-End Predictable Latent World Model Learning

Latent-Foresight is a novel end-to-end framework that jointly learns a latent tokenizer an…

Mr.Xu0 views0 discussions