quantization2 - songhan lecture
lec06 PDF File Edit View Help 1 of 1 Page 77 of 81 Edit PDF Draw Highlight Add text 118% MIT 6.5940 : TinyML and E ffi cient Deep Learning Computing https://e ffi cientml.ai If both activations and weights are binarized 66 XNOR-Net: ImageNet Classi fi cation using Binary Convolutional Neural Networks [Rastegari et al. , ECCV 2016] 1 -1 1 -1 1 1 -1 1 input weight operations memory computation = × R R + × 1 × 1 × R B + - ~32 × less ~2 × less B B xnor, popcount ~32 × less ~58 × less = -4 + popcount(1010 xnor 1101) ≪ 1 = -4 + popcount(1000) ≪ 1 = -4 + 2 = -2 y i = − n + popcount ( W i xnor x ) ≪ 1 8 -3 5 -1 5 2 0 1 = × MIT 6.5940 : TinyML and E ffi cient Deep Learning Computing https://e ffi cientml.ai Accuracy Degradation of Binarization 67 XNOR-Net: ImageNet Classi fi cation using Binary Convolutional Neural Networks [Rastegari et al. , ECCV 2016] Neural Network Quantization Bit-Width ImageNet Top-1 Accuracy W A Delta AlexNet BWN 1 32 0.2% BNN 1 1 -28.7% XNOR-Net 1 1 -12.4% GoogleNet BWN
lec06pdf Page 1 of 810 Song Han shared this file. Want to do more with it? MIT 6.5940: TinyML and Efficient Deep Learning Computinghttps://efficientml.aiEfficientML.ai Lecture 06QuantizationPart IISong HanAssociate Professor, MITDistinguished Scientist, NVIDIA@SongHan_MIT MIT 6.5940: TinyML and Efficient Deep Learning Computinghttps://efficientml.aiLecture PlanToday we will:1.Review Linear Quantization.2.Introduce Post-Training Quantization (PTQ) that quantizes a floating-point neural network model, including: channel quantization, group quantization, and range clipping.3.Introduce…
saved by
related reading
- 2112.10769arxiv.org
- lec05.pdfdropbox.com
- Integer Quantization for Deep Learning Inference: Principles and Empirical Evaluationarxiv.org
- A Visual Guide to Quantization - by Maarten Grootendorstnewsletter.maartengrootendorst.com
- [1806.08342] Quantizing deep convolutional networks for efficient inference: A whitepaperarxiv.org
- [2106.08295] A White Paper on Neural Network Quantizationarxiv.org
- The 4-bitter Lesson | humans&humansand.ai
- Quantization from the ground upngrok.com
- A Guide to Quantization in LLMs | Symbl.aisymbl.ai
- Quantization · Hugging Facehuggingface.co
- Efficient LLM inferencefinbarrtimbers.substack.com
- On neural scaling and the quanta hypothesisericjmichaud.com