AI Wiki — Knowledge Base

Sakana AI 社の最新 AI「Fugu Max」に、NVIDIA のオープンモデル「Nemotron」ファミリーが組み込まれています🐡✨オープンモデルの連携がもたらす革新的なパフォーマンス、ぜひ引用元のブログでチェックしてみてください👇

Sakana AI 社の最新 AI「Fugu Max」に、NVIDIA のオープンモデル「Nemotron」ファミリーが組み込まれています🐡✨オープンモデルの連携がもたらす革新的なパフォーマンス、ぜひ引用元のブログでチェックしてみてください👇 Introducing Fugu Max and Fugu Ultra v2: the next evolution of Sakana Fugu’s mu

model-releasesdavid-ha--x

11 Sept 2026

model releases

RiVaT-Fuse: Reliability-Calibrated Variational Tensor Fusion for Multimodal Prediction under Modality Uncertainty

Robust Multimodal Sentiment Analysis with Incomplete Modalities via Semantic-aware Completeness based Reconstruction

Running Qwen3.8-27B-Q4 at max context on a 32 GB GPU while avoiding kvcache quantization

Safe to Stop? Risk-Constrained Stopping for Sequential Clinical Diagnosis Agents

Sakana AI 社の最新 AI「Fugu Max」に、NVIDIA のオープンモデル「Nemotron」ファミリーが組み込まれています🐡✨オープンモデルの連携がもたらす革新的なパフォーマンス、ぜひ引用元のブログでチェックしてみてください👇

Same Day, Same Story; One Day Ahead, a Different Signal: The Dual Validity of Financial Sentiment

SEAR: Segment-Evidence-Aware Routing for Weak-to-Strong Multilingual Speech MCQ

SegCol Challenge: Semantic Segmentation for Tools and Fold Edges in Colonoscopy data

Self-Supervised Cardiac Phase Detection via Single-Parameter Latent Orbits

SGA: Plug&Play Geometric Verification for Educational Video Synthesis

Shedding Light: A Benchmark for Evaluating Lighting Understanding in Generative Image Models

Smoothing the Score Function to Enhance Generalization in Diffusion Models

SoK: Privacy Attacks on Machine Learning via Explainable AI

Solving Few-Shot Multiobjective Multitask Optimization via Iterative Sequential Transfer

Spec-Harness: Measuring and Improving Behavioral Adequacy of LLM-Synthesized Formal Specifications

Spectral Adapters for Segment Anything Model-based Segmentation of Colorectal Liver Metastases in Computed Tomography

Spomin - Live KV cache compaction (Experimental for Qwen)

Story Imprinting: AI Assistants Absorb Traits from Human Characters They Resemble

Structural priors for data-efficient language learning

Structurally Speaking: Motif-Oriented Graph Captioning through Bidirectional Graph-Text Translation

Structured Transforms for Low-Overhead Quantization of Language Models

Studying Without a Syllabus: Task-Agnostic Environment Preprocessing

Subgroup Membership Inference Audits of Differentially Private Synthetic Text

Symmetry-aware super-resolution of crystal orientation maps via invariant latent-space learning

SynCo: Synthetic Community-Aware Attributed Graph Generator for Graph Neural Network Benchmarking

T1: Terminal Agent Reinforcement Learning for Long-Horizon Tasks

Terminal Bench v4 scores

Thank you, Ollama. DeepSeek v4.1 Flash is the best.

The Era by Eon Benchmark: A Generated Enterprise Estate with Exact Ground Truth for Benchmarking LLM Agents

The Platonic brain bridge hypothesis: human brain networks as an architectural prior for omni models

The Truth Was Never Gone: Perfect Aliasing in Compliant-Context Truth Probes

Think Before You Link: Rarity, Reasoning, and Retrieval in Multilingual Entity Linking

Threat intelligence report: Anthropic says it disrupted a Yemen-based guided weapons engineering cell using Claude to develop guidance software for missiles (Bloomberg)

Towards Reliable Medical LLMs: Benchmarking and Enhancing Confidence Estimation of Large Language Models in Medical Consultation

TransClean: A Benchmark for Detecting and Extracting Clean Translations from Large Language Model Outputs

Trust Me, I'm Your Developer: Self-Issued Authentication in Large Language Models

Two-Parameter Flow Map Learning for Continuous-Time Diffeomorphic Image Registration

UniH^3: Unifying Hierarchical Homogeneity and Heterogeneity for All-in-One Medical Image Restoration

US-hosted, with Zero Data Retention enabled by default https://lmstudio.ai/models/deepseek-v4.1-flash

Vision Transformer-Based Multi-Level Feature Fusion for Multi-Label Sewer Defect Classification

Watermarks Without Verification: AI Text Watermarking After the EU AI Act

We discuss AI alignment and p(Doom) mid pod Also data quality vs performance, DeepSeek Flash v4.1, Rentosertib & omg a whole lot more Poor @…

Weighted Empirical Risk Minimization for Machine Learning under Long-Range Dependence: Exact Pathwise Rates and Learning-Error Geometry

When do cheap embeddings beat protein language models? A theoretically-grounded hashing sketch for biological sequence classification

When is Test-Time Adaptation Identifiable From Unlabeled Evidence?

When Synthetic Data Hurts: On Catastrophic Forgetting in Skill Retrieval for LLM Agents

Which Tokens Should SFT Actually Learn? A Token-Trimming Perspective on Mathematical Reasoning

Whisper-Based Speech Transcription from Videos Across Multiple Languages for Cross-Cultural Understanding