Tagged articles
Model Acceleration
25 articles · Page 1 of 1
RA-CFGCache introduces a risk-aligned caching framework for classifier-free guided diffusion models that jointly models cross-branch error alignment and timestep-dependent error propagation, achieving superior efficiency-fidelity trade-offs on FLUX.1-dev, Wan2.1, and CogVideoX while remaining compatible with existing cache estimators.
CachingClassifier-Free GuidanceGenerative AI
0 likes · 19 min read

OpenVDN's VDN-H3 achieves 75x speedup over MiniMax H3 by combining sliding-window softmax attention with a novel frame-level Video Delta Attention, generating 14.4 seconds of 768p video in 11.23 seconds on 8 NVIDIA B200 GPUs while preserving visual quality.
DMD2 distillationMiniMax H3Model Acceleration
0 likes · 6 min read

Unsloth Desktop, an open‑source native app for macOS, Windows and Linux, brings local large‑model inference and training into a graphical interface, delivering up to 2× faster training, 70% lower VRAM usage, and integrated features such as image/video generation, tool calling, and API compatibility while eliminating the need for Python‑based setup.
AI toolsModel AccelerationUnsloth Desktop
0 likes · 6 min read

FlashAR transforms pretrained autoregressive image models into highly parallel generators, delivering up to 22.9× end-to-end speedup while using just 0.05% of the original training data and preserving generation quality, thanks to intermediate branching, a learnable fusion gate, and a two-stage adaptation process.
FlashAR+Model Accelerationautoregressive generation
0 likes · 10 min read

OpenClaw 3.12 brings a complete UI redesign, a first‑class Ollama onboarding flow, a Fast Mode API switch, Gemini‑based multimodal memory, a new sessions_yield routing feature, robust failover mechanisms and dozens of security patches, all aimed at improving stability and performance for both casual users and large‑scale automation workloads.
AI agentsFast modeModel Acceleration
0 likes · 7 min read


AIWalkerMar 10, 2026 · Artificial Intelligence
The paper introduces MIGM-Shortcut, a self‑supervised method that learns controlled latent‑state dynamics to bypass redundant bidirectional attention in Masked Image Generation Models, achieving over 4× speed‑up on state‑of‑the‑art multimodal diffusion models like Lumina‑DiMOO while preserving image quality.
AILatent DynamicsMIGM
0 likes · 8 min read

The article analyzes the severe inference bottlenecks of large language models, CNNs, and recommendation systems and presents a suite of research‑driven accelerations—including token‑level pipeline parallelism (HPipe), KV‑cache clustering (ClusterAttn), quantization (QoKV), heterogeneous edge frameworks (DeepZoning, PICO), delay‑aware edge‑cloud scheduling (DECC), and operator choreography (RACE)—validated on real‑world industrial workloads.
AI inferenceModel Accelerationedge AI
0 likes · 16 min read

The talk outlines the AI+ era's digital ecosystem, traces the evolution from traditional AI to Agentic AI, examines emerging AI Agent technologies, and shares concrete enterprise‑level development practices, frameworks, and governance strategies for financial industry deployments.
AI agentsAgentic AIModel Acceleration
0 likes · 18 min read


AntTechSep 25, 2025 · Artificial Intelligence
The ICCV 2025 live session will deep‑dive into two cutting‑edge papers—PixTrace with CopyNCE for precise image copy detection and Skip‑Vision for dramatically faster training and inference of vision‑language models—showcasing their methods, results, and real‑world impact.
Computer VisionICCV 2025Model Acceleration
0 likes · 5 min read

TwigVLM introduces a lightweight “twig” module that prunes visual tokens early and enables self‑speculative decoding, achieving up to 154% speedup on long‑text generation while preserving 96% of original LVLM accuracy, as demonstrated on LLaVA‑1.5‑7B and other benchmarks.
LVLMModel Accelerationmultimodal AI
0 likes · 14 min read

FLUX-Lightning, introduced by PaddleMIX, combines phased consistency distillation, adversarial learning, distribution‑matching distillation, and reflow loss to reduce diffusion model inference to just four steps while preserving image quality, and leverages the CINN compiler to achieve over 30% speed gains on A800 GPUs, surpassing existing SOTA acceleration methods.
AI inferenceCINNFlux
0 likes · 21 min read

This article walks through TensorRT's core optimization features, auxiliary debugging tools, and a step‑by‑step SMPLer‑X case study, showing how graph simplification, mixed‑precision, and engine generation cut inference latency to roughly 22‑29% of the original runtime.
GPU InferenceModel AccelerationONNX
0 likes · 6 min read

This article reviews the SpargeAttn paper, describing how a training‑free sparse attention mechanism achieves 4‑7× inference speedup across language, video, and image models while preserving end‑to‑end accuracy, and outlines its challenges, algorithmic solutions, implementation details, and experimental results.
GPU optimizationModel AccelerationQuantized Inference
0 likes · 7 min read

The talk outlines how Baidu’s Baige AIAK‑LLM suite tackles the exploding compute demands of trillion‑parameter models by boosting Model FLOPS Utilization through advanced parallelism, memory‑saving recompute, zero‑offload, adaptive scheduling, and cross‑chip orchestration, delivering 30‑60% training and inference speedups and a unified cloud product.
AI infrastructureBaiduInference Optimization
0 likes · 25 min read

The talk by Tencent Games AI lead Qiu Dongyang explores the opportunities large models bring to game NPCs, showcases the "JueZhi Anuan" intelligent NPC prototype, discusses core challenges and acceleration techniques, and outlines future plans and compliance considerations for AI‑driven characters.
AIModel AccelerationNPC
0 likes · 2 min read

Our benchmark of FlashAttention2, OneFlow, DeepCache, Stable‑Fast, and LCM‑LoRA on Stable Diffusion models shows that DeepCache combined with PyTorch 2.2 consistently cuts inference time by 40‑50% with minimal code changes, while OneFlow offers 20‑40% speedups when compatible, making DeepCache the recommended default acceleration.
DeepCacheFlashAttention2LCM-LoRA
0 likes · 10 min read

PP-LCNet is a lightweight convolutional neural network designed for Intel CPUs that leverages MKLDNN acceleration, H‑Swish activation, selective SE modules, larger kernels, and expanded fully‑connected layers to achieve higher accuracy without increasing inference latency across image classification, detection, and segmentation tasks.
CPU optimizationMKLDNNModel Acceleration
0 likes · 25 min read

Alibaba Mom’s AI‑driven content risk platform has evolved from simple rule‑matching to a data‑centric, serverless architecture that integrates large‑model acceleration, decision‑tree compilation, high‑throughput vector retrieval and elastic word‑matching, delivering sub‑100 ms text and sub‑1 s image moderation while remaining stable during peak promotional traffic.
AIContent moderationDevOps
0 likes · 25 min read

This article explains how PAI‑Blade uses compiler optimizations, TorchDynamo, MHLO conversion, and aggressive kernel fusion to accelerate PyTorch training, provides simple two‑line integration code, showcases benchmark results on A10 and A100 GPUs, and details deployment steps on PAI‑DSW.
BladeDISCGPU optimizationModel Acceleration
0 likes · 8 min read

This article summarizes a collection of newly released open‑source AI projects covering natural‑language processing, multimodal processing, intelligent agents, recommendation systems, and model training acceleration, providing brief descriptions, key capabilities, and links to their repositories.
AIModel Accelerationlarge language models
0 likes · 9 min read

This article details HuoLala's AI‑driven intelligent quality inspection system, covering its NLP‑based semantic understanding pipeline, data denoising, confidence learning, contrastive learning, model acceleration techniques such as pruning, knowledge distillation, quantization, and interpretability methods to improve coverage, recall and risk detection.
Contrastive LearningModel AccelerationNLP
0 likes · 23 min read

PaddleNLP v2.1 introduces an industrial‑grade Taskflow for eight NLP scenarios, a three‑line few‑shot learning paradigm that boosts small‑sample performance, and a FasterTransformer‑based inference engine that delivers up to 28‑fold speedup for text generation, all backed by extensive model and algorithm integrations.
Few-Shot LearningModel AccelerationNLP
0 likes · 7 min read

This lecture outlines the key challenges of deep learning in retail—growing model size, speed, and energy consumption—and presents a comprehensive acceleration framework covering algorithmic optimizations like network design, pruning, and hardware acceleration, with practical examples such as MobileNet, model compression, and edge deployment.
Model Accelerationdeep learninghardware optimization
0 likes · 15 min read

Deep learning models often suffer from slow training and deployment due to their size, but a range of advanced acceleration methods—including model architecture optimization, pruning, quantization, knowledge distillation, and distributed training techniques—can dramatically improve speed and efficiency while maintaining performance.
Model Accelerationdeep learningdistributed training
0 likes · 14 min read

This article reviews why deep neural networks are over‑parameterized, outlines the challenges of deploying them on mobile and embedded devices, and presents six major strategies—pruning, low‑rank approximation, filter selection, quantization, knowledge distillation, and novel architecture design—to accelerate and compress models while preserving performance.
Model Accelerationdeep learningknowledge distillation
0 likes · 11 min read
