What’s new in PyTorch 1.3

on YouTube ↗12m

" summary

PyTorch 1.3 adds explicit named tensors to attach semantics to tensor dims, a TorchScript-based mobile workflow that enables 60 FPS MobileNet on iPhone for on-device inference, and a quantization toolkit offering dynamic / post-training / quantization-aware paths so teams trade off accuracy vs latency. Backends integrate MKL-DNN, cuDNN, XLA and Glow for faster kernels, and gradient-attribution + TensorBoard hooks ship to inspect and visualize model decisions.

AI-written summary · about

" chapters (5)

1. 0:00Introduction
2. 0:24What is PyTorch
3. 8:05Mobile
4. 9:17Quantization
5. 10:25Conclusion

" related talks

• 08mPyTorch MobileDavid Reiss · PyTorch Developer Day 2020
• 09mPyTorch Mobile: Past, present and 2.x futureRAZIEL ALVAREZ GUEVARA · PyTorch Conference 2022
• 54mKeynote: PyTorch 2.1 Technical Deep DiveJoe Spisak, Peng Wu, Mergen Nachin, Mark Saroufim, Yanan Cao, Will Constable, Mario Lezcano · PyTorch Conference 2023
• 06mPyTorch Mobile and Android Neural Networks APIOli Gaymond · PyTorch Developer Day 2020
• 06mKeynote: PyTorch Lightning: Powering the GenAI Revolution from Research to the...William Falcon · PyTorch Conference 2023
• 35mClient Side Deep Learning Optimization with PyTorchShane Caldwell, Tyler Kirby · Strange Loop 2021
• 16mLightning Talk: Efficient Inference at the Edge: Performance You Need at the Lowest...Felix Baum · PyTorch Conference 2023
• 14mOptimize your TensorFlow Lite modelsEthan Kim, Yunlu Li · Google I/O 2021