The paper introduces Hardware‑Aware FP4 FlashAttention‑4, which optimizes attention mechanisms for NVIDIA’s Blackwell 4‑bit floating‑point (FP4) tensor cores. By employing Direct‑P for noncausal inference and a causal path that forwards quantized scores into the backward pass, the method achieves up to 2.13× the bfloat16 forward throughput on an NVIDIA GB200. The causal approach also reconstructs probabilities from saved quantized queries and keys, using 8‑bit floating‑point (FP8) gradients to accelerate a full single‑GPU 8‑billion‑parameter update by up to 1.14×, while distributed training with FP8 probabilities and values shows divergent trajectories compared to tested MXFP4 setups.
arXiv:2603. 00040v3 Announce Type: replace-cross Abstract: Achieving reliable 4-bit attention is a prerequisite for end-to-end FP4 computation on emerging FP4-capable GPUs, yet attention remains the main obstacle due to FP4's tiny dynamic range and attention's heavy-tailed activations.
By Peiyuan Zhang, Matthew Noto, Wenxuan Tan, Chengquan Jiang, Will Lin, Wei Zhou, Hao Zhang
arXiv:2601. 22813v2 Announce Type: replace Abstract: The NVFP4 lower-precision format, supported in hardware by NVIDIA Blackwell GPUs, promises to allow, for the first time, end-to-end fully-quantized pre-training of massive models such as LLMs.
By Andrei Panferov, Erik Schultheis, Soroush Tabesh, Dan Alistarh
arXiv:2606. 15682v1 Announce Type: new Abstract: Large Reasoning Models (LRMs) achieve strong problem-solving through long chain-of-thought, but their deployment is constrained by the high cost of full-precision inference and growing KV cache footprints.
By Janghwan Lee, Sihwa Lee, Jinseok Kim, Yongjik Kim, Jieun Lim, Jinwook Oh, Jungwook Choi
arXiv:2605. 09825v4 Announce Type: replace-cross Abstract: Why does full-pipeline FP4 training of large language models often diverge, even when forward activations and activation gradients remain stable?
By Musa Cim, Sarthak Arora, Poovaiah Palangappa, Miro Hodak, Ravi Dwivedula, Meena Arunachalam, Mahmut Taylan Kandemir
arXiv:2607. 24953v1 Announce Type: cross Abstract: Reducing training precision is a key lever for improving the e ciency of large language model (LLM) training, but pushing beyond FP8 to 4-bit oating point (FP4) remains challenging due to instability during optimization.
By Mehdi Rahimifar, Amin Darabi, Mehran Taghian Jazi, Xing Huang, Yao Wang, Zhijun Tu, Yufei Cui, Yunke Peng, Hongliang Li
Stable 4-bit floating-point (FP4) pretraining is difficult because the E2M1 payload represents only a narrow range of magnitudes. NVIDIA's Transformer Engine \nv{} recipe addresses this with current-tensor scaling, a randomized Hadamard transform (RHT), and bfloat16 (BF16) final layers, adding work outside the FP4 matrix multiplications.
arXiv:2603. 10444v2 Announce Type: replace-cross Abstract: FP4 training promises substantial memory and compute savings for large language models, but remains fragile because blockwise quantization is dictated by extreme activation magnitudes, which inflate dynamic range and compress long-tail signals.
By Hengjie Cao, Zhendong Huang, Mengyi Chen, Yifeng Yang, Fang Dong, Anrui Chen, Ruijun Huang, Xin Zhang, Mingzhi Dong, Yujiang Wang, Jinlong Hou, Qin Lv, Robert P. Dick, Yuan Cheng, Tun Lu, Fan Yang, Yixuan Chen, Li Shang
arXiv:2506. 01969v3 Announce Type: replace-cross Abstract: Efficient inference of Multi-Head Latent Attention (MLA) is challenged by deploying the DeepSeek-R1 671B model on a single Multi-GPU server.
By Pengcuo Dege, Qiuming Luo, Rui Mao, Chang Kong
arXiv:2606. 20381v1 Announce Type: new Abstract: FP4 training promises substantial reductions in memory and computation cost for LLM pretraining, yet current FP4 hardware paths and recipes, including NVIDIA Blackwell/Rubin-class systems and AMD MI350-series GPUs, remain centered on E2M1 data elements.
By Qian Zhao, Kunlong Chen, Changxin Tian, Zhonghui Jiang, Haitao Zhang, Chaofan Yu, Peijie Jiang, Mingliang Gong, Jia Liu, Ziqi Liu, Zhiqiang Zhang, Jun Zhou
The paper introduces a new 4‑bit floating‑point (FP4) pretraining approach that pairs E2M1 payloads with unsigned E5M3 block scales, enabling periodic tensor scaling and selective stochastic rounding while eliminating the randomized Hadamard transform. Using this method, the authors pretrained a Nemotron‑H 8B model on nearly 190 billion tokens, achieving lower training and validation losses compared to NVIDIA’s Transformer Engine. The approach also improves inference performance and demonstrates a 21.2 % increase in token throughput when certain optimizations are removed.
By Robert Hu, Carlo Luschi, Paul Balanca
arXiv:2608. 19758v1 Announce Type: new Abstract: Long-context modeling is a pivotal capability for Large Language Models, yet the quadratic complexity of attention remains a critical bottleneck, particularly during the compute-intensive prefilling phase.
By Qihang Fan, Huaibo Huang, Zhiying Wu, Bingning Wang, Ran He