AI Wiki — Knowledge Base

Finally got qwen 3.8 q4 k_M 27b runnin slick 40tkps load at 240k inf 4qnl for stable agentic coding on a 7900xtx. I have been finally able to give hermes a task and come back to results. Built a panel to manage my inference servers from hermes.

Screenshot of heavy load. I did offload vision to a spare 2060 instead of local ram. The control panel was generated with qwen3.8-27b davidau q4k_m max mtp. Made the panel from scratch over a week, ba

model-releasesr-localllama

19 Sept 2026

Search: “reasoning”

Teaching and Evaluating LLMs to Reason About Polymer Design Related Tasks

Reason to Play: Behavioral and Brain Alignment Between Frontier LRMs and Human Game Learners

Large Language Models are Universal Reasoners for Visual Generation

SCPRM: A Schema-aware Cumulative Process Reward Model for Knowledge Graph Question Answering

ReMedi: Reasoner for Medical Clinical Prediction

SSL-R1: Self-Supervised Visual Reinforcement Post-Training for Multimodal Large Language Models

A-MAR: Agent-based Multimodal Art Retrieval for Fine-Grained Artwork Understanding

Does Self-Consistency Improve the Recall of Encyclopedic Knowledge?

Probabilistic Programs of Thought

QuantSightBench: Evaluating LLM Quantitative Forecasting with Prediction Intervals

ChartNet: A Million-Scale, High-Quality Multimodal Dataset for Robust Chart Understanding

PILOT: Planning via Internalized Latent Optimization Trajectories for Large Language Models

Can LLMs Reason About Attention? Towards Zero-Shot Analysis of Multimodal Classroom Behavior

RecaLLM: Addressing the Lost-in-Thought Phenomenon with Explicit In-Context Retrieval

Efficient PRM Training Data Synthesis via Formal Verification

SpatialMosaic: A Multiview VLM Dataset for Partial Visibility

Training Generalist Navigation Policies via Action Self-Distillation

MolP2SBench: Do LLMs Really Reason from Molecular Properties to Structure?

State-OPD: State-Grounded On-Policy Distillation for Multi-Turn Agents

What setup do you use, which harness? and how do you customize it to run qwen 3.8 flash next swift with a high context length and a reasonable amount of tok/sec?

Do Higher-Order Models Win for Higher-Order Reasons? Rethinking Performance Gains in Hypergraph Learning

Learning to Reason with Compressed Context: Ground-Truth-Free Adaptation of OmniLLMs via Self-Distillation

Learning to Reason with Persistent Object States for Video Instance Segmentation

How Children Design and Reason about Trustworthy AI Chatbots

SHRAV: State-Hypothesis-Reason-Action-Verify Framework for Physical Modeling and Inverse Design

Construting Reverse Thinking: Developing Large Language Models' Reverse Thingking Ability

VeriFuse: Bounded Vision-Language Arbitration and Reason-Guided Refinement for Cooperative 3D Perception

Finally got qwen 3.8 q4 k_M 27b runnin slick 40tkps load at 240k inf 4qnl for stable agentic coding on a 7900xtx. I have been finally able to give hermes a task and come back to results. Built a panel to manage my inference servers from hermes.

WordPolo: Evaluating Language Models Through Iterative Semantic Feedback

SceneBench: A Hierarchical Benchmark for Vision-Language Understanding of 3D Scenes

VeriDx: Earning the Right to Diagnose with Disease-Centric Verification

Beyond Similarity: Foundation Models as an Efficient Backbone for Training-Free Composed Video Retrieval

Reactivating Test-Time Scaling for Plane Geometry Problem Solving

Think, Look, and Revise: Inconsistency-Aware Visual Self-Correction in MLLMs

Triggering Chain-of-Thought via Latent Feature Interventions in Large Language Models

Wrong Prediction, Right Answer: Recovering Evidence from Collapsed LLM Sequence Scores

Qwen3.8-Flash-Next-NVFP4 vs Qwen3.8-27B-FP Test Results

CritICL: Inference-Time Weak-to-Strong Generalization from Small Language Model Failure Modes

Learning to Predict, Discover, and Reason in High-Dimensional Event Sequences

Prefix Sliding for efficient test-time scaling

Can a Dynamic Internal Field Govern a Transformer's Cognition? Certifiability, not Superiority, in Homeostatic Compute Control

Effects of Theory of Mind and Prosocial Beliefs on Steering Human-Aligned Behaviors of LLMs in Ultimatum Games

VGI-BENCH: Probing Visual Intelligence in Video Generation Models

CrossView: Can Vision-Language Models Reason Across Cameras?

Towards Reasonable Molecular Structure Elucidation from Infrared Spectroscopy with Chemical Feedback

How Do VLMs Behave When Blind or Misled? Behavioral Evaluation of VLMs on Scientific Figures

StreamReason-Bench: Can Large Language Models Reason about Event-Time Stream-Processing Semantics?

Think Deep, Speak Once: Relit, A Recursive Latent Implicit Transformer Framework