#paddlepaddle

The latest AI and developer news about #paddlepaddle, with the original source and a short summary.

Virtual Head Attention Reduces KV Cache by 4x and Improves Quality Over GQA

VHA uses linear operations to halve physical query heads, achieving 4x KV cache reduction and +0.70% quality improvement over MLA on a 30B MoE model.

·2026.10.24 00:00

FlashMask-3 Achieves 45-141% TFLOPs Gain Over FlashMask on H100 GPUs

The new Hopper kernel and distributed extension improve throughput by 1.80x to 3.71x over Megatron-Core baselines at context-parallel degrees 4-32.

·2·2026.10.20 23:00

PickHPD-Parsing Achieves SoTA in Document Parsing

PaddlePaddle has released HPD-Parsing, a 1B document parsing model based on hierarchical parallel decoding, achieving the top score of 94.91% on OmniDocBench v1.6.

·1·2026.07.23 16:00

Hugging Face begins supporting PaddlePaddle models

The Hugging Face Hub now supports models from PaddlePaddle, Baidu's open-source deep learning framework.

·1·2023.01.17 09:00