HX
Haocheng Xi
cs.LGcs.CVcs.AIcs.CLcs.PF
On Valency
published · living versionsW_472urbcs·v1 · currentpublished
FrontierCS 2: A Benchmark for Computer Science R&D
with Qiuyang Mang, Wenhao Chai, Runyuan He, Dongruixuan Li +56
1 version
Preprints & journals
22 papers in the corpus · 2024–2026FrontierCS 2: A Benchmark for Computer Science R&D2610.29932v1 · Qiuyang Mang, Wenhao Chai, Runyuan He et al.2026 · 0 citationsarXivon Valency
LeapQuant: Efficient Linear Attention with Accurate Recurrent State Quantization2609.38166v1 · Yi Pan, Haocheng Xi, Kan Zhu et al.2026 · 0 citationsarXiv
Video DeltaNet: A Video-Native Hybrid Attention for Livestream Video Generation2609.20744v3 · Haocheng Xi, Yiming Xie, Hexu Zhao et al.2026 · 0 citationsarXiv
VC-Attention: Value Smoothing and Softmax Casting for Low-bit Attention2609.15810v1 · Xingyang Li, Dongyun Zou, Shining Zhang et al.2026 · 0 citationsarXiv
FourTune: Towards Fully 4-Bit Efficient Post-Training for Diffusion Models2607.05711v1 · Bowen Xue, Zihan Min, Xingyang Li et al.2026 · 0 citationsarXiv
LoSA: Locality Aware Sparse Attention for Block-Wise Diffusion Language Models2604.12056v1 · Haocheng Xi, Harman Singh, Yuezhou Hu et al.2026 · 0 citationsarXiv
SVG-EAR: Parameter-Free Linear Compensation for Sparse Video Generation via Error-aware Routing2603.08982v2 · Xuanyi Zhou, Qiuyang Mang, Shuo Yang et al.2026 · 0 citationsarXiv
SpargeAttention2: Trainable Sparse Attention via Hybrid Top-k+Top-p Masking and Distillation Fine-Tuning2602.13515v2 · Jintao Zhang, Kai Jiang, Chendong Xiang et al.2026 · 0 citationsarXiv
Quant VideoGen: Auto-Regressive Long Video Generation via 2-Bit KV-Cache Quantization2602.02958v5 · Haocheng Xi, Shuo Yang, Yilong Zhao et al.2026 · 0 citationsarXiv
Residual Context Diffusion Language Models2601.22954v2 · Yuezhou Hu, Harman Singh, Monishwaran Maheswaran et al.2026 · 0 citationsarXiv
Jet-RL: Enabling On-Policy FP8 Reinforcement Learning with Unified Training and Rollout Precision Flow2601.14243v2 · Haocheng Xi, Charlie Ruan, Peiyuan Liao et al.2026 · 0 citationsarXiv
DC-Gen: Post-Training Diffusion Acceleration with Deeply Compressed Latent Space2509.25180v3 · Wenkun He, Yuchao Gu, Junyu Chen et al.2025 · 1 citationarXiv
DC-VideoGen: Efficient Video Generation with Deep Compression Video Autoencoder2509.25182v1 · Junyu Chen, Wenkun He, Yuchao Gu et al.2025 · 0 citationsarXiv
Jet-Nemotron: Efficient Language Model with Post Neural Architecture Search2508.15884v3 · Yuxian Gu, Qinghao Hu, Shang Yang et al.2025 · 1 citationarXiv
Radial Attention: Sparse Attention with Energy Decay for Long Video Generation2506.19852v2 · Xingyang Li, Muyang Li, Tianle Cai et al.2025 · 0 citationsarXiv
Sparse VideoGen2: Accelerate Video Generation with Sparse Attention via Semantic-Aware Permutation2505.18875v5 · Shuo Yang, Haocheng Xi, Yilong Zhao et al.2025 · 7 citationsarXiv
Oscillation-Reduced MXFP4 Training for Vision Transformers2502.20853v2 · Yuxiang Chen, Haocheng Xi, Jun Zhu et al.2025 · 0 citationsProceedings of the 42nd International Conference on Machine Learning, PMLR 267, 2025 (ICML 2025)
SpargeAttention: Accurate and Training-free Sparse Attention Accelerating Any Model Inference2502.18137v8 · Jintao Zhang, Chendong Xiang, Haofeng Huang et al.2025 · 0 citationsProceedings of the 42 nd International Conference on Machine Learning, PMLR 267, 2025 (ICML 2025)
Sparse VideoGen: Accelerating Video Diffusion Transformers with Spatial-Temporal Sparsity2502.01776v2 · Haocheng Xi, Shuo Yang, Yilong Zhao et al.2025 · 1 citationarXiv
Humanity's Last Exam2501.14249v12 · Long Phan, Alice Gatti, Ziwen Han et al.2025 · 61 citationsarXiv
NVILA: Efficient Frontier Visual Language Models2412.04468v3 · Zhijian Liu, Ligeng Zhu, Baifeng Shi et al.2024 · 45 citationsarXiv
T-Rex: Text-assisted Retrosynthesis Prediction2401.14637v1 · Yifeng Liu, Hanwen Xu, Tangqi Fang et al.2024 · 1 citationarXiv
Career total: 25 works. 22 are in this corpus.
Profile built from the corpus for this byline.
Author records are still filling in while the Hub is in alpha. If this is your page, you'll be able to claim it soon. Spot a mistake? Tell us.