AI Wiki — Knowledge Base

This is pleasant. SDXL/DMD-2 images, SEEDVR2, LTX-2.3, pieced together with Shotcut. Overall the whole thing took a couple days, just tweaking moments in Comfy, getting about 90 images together, cutting it down, ended up running 30 through LTX on a 3060 12GB/64GB - might get some vocals~

This post documents a creative project combining multiple AI image and video generation tools—SDXL, DMD-2, SEEDVR2, and LTX-2.3—with video editing software Shotcut to create a multi-day production. Th

local-air-stablediffusion

3 Jun 2026

Search: “hardware”

CourseTimeQA: A Lecture-Video Benchmark and a Latency-Constrained Cross-Modal Fusion Method for Timestamped QA

Dynamic Short Convolutions Improve Transformers

Evaluating LLMs' Effectiveness on Real-World Consumer Device Repair Questions

Fixed-Time Dynamic Landing of Quadrotors using Adaptive Unscented Kalman Filtering and Nonlinear Model Predictive Control

FlashMLA-ETAP: Efficient Transpose Attention Pipeline for Accelerating MLA Inference on NVIDIA H20 GPUs

KForge: LLM-Driven Cross-Platform Kernel Generation for AI Accelerators

Latent Activation Editing: Inference-Time Refinement of Learned Policies for Safer Multirobot Navigation

Link Prediction or Perdition: the Seeds of Instability in Knowledge Graph Embeddings

LiveBand: Live Accompaniment Generation in the Audio Domain

Multi-Segment Attention: Enabling Efficient KV-Cache Management for Faster Large Language Model Serving

NetKV: Network-Aware Decode Instance Selection for Disaggregated LLM Inference

OASIS: Outlier-Aware LUT-Based GEMM with Dual-Side Quantization for LLM Inference Acceleration

Optimizing Explicit Unit-Distance Lower-Bound Certificates

PrimeSVT: An Automated Memory-aware Pruning Framework with Prioritized Compression Policy for Spiking Vision Transformers

PSViT: A Methodology for Structurally Pruning Spiking Vision Transformers

QUIVER: Quantum-Informed Views for Enhanced Representations in Large ML Models

Qwen3.6-35B-A3B on 2× GTX 1080 Ti with Ollama: ~20 tok/s + 3 gotchas (driver 570+, cuda_v12 for Pascal, quant fit on 22GB)

Reproducibility is the New Copyleft: Defining AGI-oriented Reproducible Builds

SPADE: Sketch-guided Path Planning Augmented with Diffusion Experts

This is pleasant. SDXL/DMD-2 images, SEEDVR2, LTX-2.3, pieced together with Shotcut. Overall the whole thing took a couple days, just tweaking moments in Comfy, getting about 90 images together, cutting it down, ended up running 30 through LTX on a 3060 12GB/64GB - might get some vocals~

Toward Gripper-Integrated Active Electrosense for Pre-Contact Sensing in Underwater Soft Grippers

v0.30.2

v0.30.4: llama-server: fix gemma4 patch wiring (#16477)

WISE-HAR: A Generalizable Ensemble Deep Learning Framework for WiFi-Based Human Activity Recognition

A Four-Tier Communication Architecture and Sim-to-Real Validation of a Graphical Open-Source Platform for Robotic Engineering Education

Adaptive PD Gains for Energy-Conscious Control in Physical Human-Robot Interaction

AI-PROPELLER: Warehouse-Scale Interprocedural Code Layout Optimization with AlphaEvolve

All Models are Wrong, Knowing Where is Useful: On Model Uncertainty in Reinforcement Learning

b9466

b9470

b9474

b9483

Beyond Augmentation: Score-Guided Pathological Prior for EEG-based Depression Detection

Beyond Pure Sampling: Hybrid Optimization Mechanisms for Non-Convex Model Predictive Control

Boundary-Protection W8A8 HiFloat8 Quantization for Large-Scale Text-to-Video Diffusion Transformers

Cisco bets on AI agents to redefine workplace collaboration

Directed Distance Fields for Constant-Time Ray Queries on Gaussian Splatting

DREAM-S: Speculative Decoding with Searchable Drafting and Target-Aware Refinement for Multimodal Generation

EEG-FuseFormer: A Transformer-Driven Feature Fusion Framework for Seizure Onset Prediction

Ego-METAS: Egocentric online Multimodal Energy-efficient Temporal Action Segmentation benchmark

Embedding Semantic Risk into Distance Fields and CBFs for Online Monocular Safe Control

FedMTFI: Feature Importance Based Optimized Multi Teacher Knowledge Distillation in Heterogeneous Federated Learning Environment

Honey, I Shrunk the Arc de Triomphe!

Interpretable Policy Distillation for Power Grid Topology Control

{lambda}Split: Self-Supervised Content-Aware Spectral Unmixing for Fluorescence Microscopy

LLM Trainer: Automated Robotic Data Generation via Demonstration Augmentation using LLMs

LLM4Cov: Execution-Aware Agentic Learning for High-coverage Testbench Generation

Project SPARROW and the Future of Conservation Technology