速度超快！字节跳动开源序列推理引擎LightSeq( 五 )
可视化结果说明了 LightSeq 已经做到了极致优化 ， 大大提升了推理速度 。
传送门：
GitHub项目地址：
https://github.com/bytedance/lightseq
[1] Vaswani, Ashish, et al. ''Attention is all you need.'' Advances in neural information processing systems. 2017.
[2] Devlin, Jacob, et al. ''Bert: Pre-training of deep bidirectional transformers for language understanding.'' arXiv preprint arXiv:1810.04805 (2018).
[3] Brown, Tom B., et al. ''Language models are few-shot learners.'' arXiv preprint arXiv:2005.14165 (2020).
[4] WMT2020, http://www.statmt.org/wmt20/
[5] Li, Jiwei, Will Monroe, and Dan Jurafsky. ''A simple, fast diverse decoding algorithm for neural generation.'' arXiv preprint arXiv:1611.08562 (2016).
[6] TurboTransformers, https://github.com/Tencent/TurboTransformers
[7] FasterTransformer, https://github.com/NVIDIA/DeepLearningExamples/tree/master/FasterTransformer
[8] NVIDIA Triton Inference Server, https://github.com/triton-inference-server/server
[9] LightSeq proto, https://github.com/bytedance/lightseq/tree/master/proto
[10] LightSeq性能评测报告, https://github.com/bytedance/lightseq/blob/master/docs/performance.md
[11] LightSeq Layer Normalization, https://github.com/bytedance/lightseq/blob/master/kernels/transformerKernels.cu.cc#L269
[12] cuBLAS, https://docs.nvidia.com/cuda/cublas/index.html
[13] GPT2,''Language Models are Unsupervised Multitask Learners''
推荐阅读
- 医院|网传字节跳动又有员工送医院急救 当事人回应：已回家休息
- 字节跳动|逝者母亲辟谣字节赔2000万传言 双方暂未协商赔偿事宜：我现在很痛苦
- 猝死|28岁的工程师离世在冬夜：网传字节跳动又有员工送医院急救
- 惠普|惠普FX900 Pro M.2 2TB固态硬盘上手：解决容量和速度焦虑
- 抖音|字节跳动旗下汽水音乐完成软著登记：抖音“姊妹”音乐App就要来了
- 猝死|字节跳动被传猝死员工已离世 回应：尽力帮助其家人共渡难关
- 真我|全球首款150W快充手机曝光：realme首发 速度完胜一众高端旗舰
- 猝死|28岁员工猝死？字节跳动辟谣仍在医院抢救中 当事人妻子刚怀孕两个月
- 苹果|苹果加速开发中！iOS 16曝光：速度更快、UI改动明显
- 微软|新版任务管理器曝光Win11性能新机制：速度提升多达76%