PaceVGGT reduces VGGT inference latency by up to 5.1x on ScanNet-50 via pre-AA token pruning with a distilled Token Scorer, per-frame keep budgets, adaptive merge/prune, and feature-guided restoration, while preserving reconstruction quality on ScanNet-50 and 7-Scenes.
hub
FlashAttention-2: Faster attention with better parallelism and work partitioning
10 Pith papers cite this work. Polarity classification is still indexing.
10
Pith papers citing it
hub tools
citation-role summary
method 3
background 1
citation-polarity summary
representative citing papers
LookWhen factorizes video recognition into learning when, where, and what to compute via uniqueness-based token selection and dual-teacher distillation, achieving better accuracy-FLOPs trade-offs than baselines on multiple datasets.
A transformer-encoded spherical normalizing flow achieves state-of-the-art angular resolution for IceCube neutrino tracks and showers, improving median resolution by factors of 1.3-2.5 over B-spline likelihoods at 100 TeV and outperforming prior ML methods for muons.
RPA kernel for TPUs achieves 86% MBU in decode and 73% MFU in prefill on Llama 3 8B via tiling for ragged memory, fused pipelines, and specialized compilation for prefill/decode workloads.
KV cache compression causes task-dependent degradation in high-density reasoning due to disrupted CoT links; ShotKV mitigates this by preserving few-shot examples as indivisible semantic units through phase separation, delivering 9-18% accuracy gains and 11% latency reduction.
OmniDrop is a training-free layer-wise token pruning framework for omni-modal LLMs that uses query guidance and temporal diversity to reduce prefill latency by up to 40% and memory by 14.7% while improving benchmark scores by up to 3.58 points.
ChunkFlow achieves up to 1.28x step-time speedup and up to 49% lower peak GPU memory for DiT inference by using a first-order model to guide communication-aware chunked prefetching.
BSA-TNP is a new neural process model with KRBlocks and biased scan attention that claims to match top accuracy while scaling inference to over 1M points in under a minute on a single GPU and supporting translation invariance.
ERASE prunes 85% of vision tokens in Qwen2.5-VL-7B while retaining 89.46% accuracy, outperforming prior methods that retain only 78.1%.
Step-Audio-R1.5 applies RLHF to audio reasoning models to escape the verifiable reward trap of RLVR, preserving analytical ability while restoring prosodic naturalness and immersion in long dialogues.
citing papers explorer
-
PaceVGGT: Pre-Alternating-Attention Token Pruning for Visual Geometry Transformers
PaceVGGT reduces VGGT inference latency by up to 5.1x on ScanNet-50 via pre-AA token pruning with a distilled Token Scorer, per-frame keep budgets, adaptive merge/prune, and feature-guided restoration, while preserving reconstruction quality on ScanNet-50 and 7-Scenes.
-
LookWhen? Fast Video Recognition by Learning When, Where, and What to Compute
LookWhen factorizes video recognition into learning when, where, and what to compute via uniqueness-based token selection and dual-teacher distillation, achieving better accuracy-FLOPs trade-offs than baselines on multiple datasets.
-
Neural posterior estimation of the neutrino direction in IceCube using transformer-encoded normalizing flows on the sphere
A transformer-encoded spherical normalizing flow achieves state-of-the-art angular resolution for IceCube neutrino tracks and showers, improving median resolution by factors of 1.3-2.5 over B-spline likelihoods at 100 TeV and outperforming prior ML methods for muons.
-
Ragged Paged Attention: A High-Performance and Flexible LLM Inference Kernel for TPU
RPA kernel for TPUs achieves 86% MBU in decode and 73% MFU in prefill on Llama 3 8B via tiling for ragged memory, fused pipelines, and specialized compilation for prefill/decode workloads.
-
Semantic Integrity Matters: Benchmarking and Preserving High-Density Reasoning in KV Cache Compression
KV cache compression causes task-dependent degradation in high-density reasoning due to disrupted CoT links; ShotKV mitigates this by preserving few-shot examples as indivisible semantic units through phase separation, delivering 9-18% accuracy gains and 11% latency reduction.
-
OmniDrop: Layer-wise Token Pruning for Omni-modal LLMs via Query-Guidance
OmniDrop is a training-free layer-wise token pruning framework for omni-modal LLMs that uses query guidance and temporal diversity to reduce prefill latency by up to 40% and memory by 14.7% while improving benchmark scores by up to 3.58 points.
-
ChunkFlow: Communication-Aware Chunked Prefetching for Layerwise Offloading in Distributed Diffusion Transformer Inference
ChunkFlow achieves up to 1.28x step-time speedup and up to 49% lower peak GPU memory for DiT inference by using a first-order model to guide communication-aware chunked prefetching.
-
Scalable Spatiotemporal Inference with Biased Scan Attention Transformer Neural Processes
BSA-TNP is a new neural process model with KRBlocks and biased scan attention that claims to match top accuracy while scaling inference to over 1M points in under a minute on a single GPU and supporting translation invariance.
-
ERASE: Eliminating Redundant Visual Tokens via Adaptive Two-Stage Token Pruning
ERASE prunes 85% of vision tokens in Qwen2.5-VL-7B while retaining 89.46% accuracy, outperforming prior methods that retain only 78.1%.
-
Step-Audio-R1.5 Technical Report
Step-Audio-R1.5 applies RLHF to audio reasoning models to escape the verifiable reward trap of RLVR, preserving analytical ability while restoring prosodic naturalness and immersion in long dialogues.