Tweetovi
- Tweetovi, trenutna stranica.
- Tweetovi i odgovori
- Medijski sadržaj
Blokirali ste korisnika/cu @MaratDukhan
Jeste li sigurni da želite vidjeti te tweetove? Time nećete deblokirati korisnika/cu @MaratDukhan
-
Marat Dukhan proslijedio/la je Tweet
Pay attention to this. We’re seeing up to 3X speed ups for real world models in the tfjs WASM backend with SIMD128 enabled!https://twitter.com/v8js/status/1222944308183085058 …
Hvala. Twitter će to iskoristiti za poboljšanje vaše vremenske crte. PoništiPoništi -
New goodies from colleagues at
@GoogleAI! All#MediaPipe effects - edge detection, face detection, hair segmentation, and hand tracking - now can run inside a Web browser, powered by XNNPACK and#WebAssemblyhttps://twitter.com/googledevs/status/1222237214983090176 …
Hvala. Twitter će to iskoristiti za poboljšanje vaše vremenske crte. PoništiPoništi -
Marat Dukhan proslijedio/la je Tweet
Really excited about this model, which runs in real-time on Pixel and iPhone on our WebGL and WASM backends! Really nice work
@GreenBeanDou and@wakoan!https://twitter.com/TensorFlow/status/1214290578374041600 …
Hvala. Twitter će to iskoristiti za poboljšanje vaše vremenske crte. PoništiPoništi -
TensorFlow.js on CPU now faster with an XNNPACK-powered
#WebAssembly backend! Whopping 4-20x over previous TF.js CPU backend in pure JavaScript
, near-universal coverage, and Node.js compatibility - available right now in the Alpha releasehttps://twitter.com/TensorFlow/status/1208153137774813187 …
Hvala. Twitter će to iskoristiti za poboljšanje vaše vremenske crte. PoništiPoništi -
The micro-kernels for sparse inference on ARM64 and WebAssembly are already open-sourced in XNNPACK [http://github.com/google/XNNPACK ], and so are pre-trained sparse models [http://github.com/google-research/google-research/tree/master/fastconvnets …] [4/4]
Prikaži ovu nitHvala. Twitter će to iskoristiti za poboljšanje vaše vremenske crte. PoništiPoništi -
Our recent work [https://arxiv.org/abs/1911.09723 ] with colleagues from
@DeepMind and@GoogleAI demonstrates that with a right layout and optimizations sparse inference delivers practical and non-negligible speedups of 1.3X-2.4X on a range of MobileNet and EfficientNet models. [3/4]Prikaži ovu nitHvala. Twitter će to iskoristiti za poboljšanje vaše vremenske crte. PoništiPoništi -
Computations in sparsified models involve many multiplications by zeroes, which can be skipped in theory, but common wisdom suggested that it is impractical in software inference implementations. [2/4]
Prikaži ovu nitHvala. Twitter će to iskoristiti za poboljšanje vaše vremenske crte. PoništiPoništi -
Sparsification, or pruning of weights, in convolutional neural networks has a long history as a compression technique, and good support in deep learning frameworks, e.g. Model Optimization Toolkit in TensorFlow. [1/4]
Prikaži ovu nitHvala. Twitter će to iskoristiti za poboljšanje vaše vremenske crte. PoništiPoništi -
Marat Dukhan proslijedio/la je Tweet
“Fast Sparse ConvNets”, a collaboration w/
@GoogleAI [https://arxiv.org/abs/1911.09723 ], implements fast Sparse Matrix-Matrix Multiplication to replace dense 1x1 convolutions in MobileNet architectures. The sparse networks are 66% the size and 1.5-2x faster than their dense equivalents.pic.twitter.com/poDKMzfA4u
Prikaži ovu nitHvala. Twitter će to iskoristiti za poboljšanje vaše vremenske crte. PoništiPoništi -
Live in-browser Hand Tracking + Landmark Detection demos:http://mediapipe.page.link/web
Prikaži ovu nitHvala. Twitter će to iskoristiti za poboljšanje vaše vremenske crte. PoništiPoništi -
A preview of in-browser machine learning-based demos by our group in a great
#ChromeDevSummit talk by@fractorious and@RReverser https://youtu.be/kZrl91SPSpc?t=596 … Powered By XNNPACK, MediaPipe, and#WebAssembly (+SIMD)pic.twitter.com/oYr0Jp0jvS
Prikaži ovu nitHvala. Twitter će to iskoristiti za poboljšanje vaše vremenske crte. PoništiPoništi -
Third-generation NNPACK-family library is here, at http://www.github.com/google/XNNPACK ! This time the focus is on accelerating FP32 models in NHWC layout, and it supports both mobile and Web platforms.
Hvala. Twitter će to iskoristiti za poboljšanje vaše vremenske crte. PoništiPoništi -
Today Facebook publicly released QNNPACK, open source library for low-precision neural network computations on mobile. Caffe2+QNNPACK = 2x speedup over TFLite + support for grouped conv (CondenseNet, ShuffleNet, RexNeXt).https://code.fb.com/ml-applications/qnnpack/ …
Hvala. Twitter će to iskoristiti za poboljšanje vaše vremenske crte. PoništiPoništi -
Marat Dukhan proslijedio/la je Tweet
We've just released our new
@PyTorch tutorial on how to interface it with Numpy and use it for#DeepLearning in#manufacturing. Go and check it out!@hpcgarage@GTCSE@gtcomputing@GeorgiaTech@Wafa_louhichi@DocFasthttps://github.com/hpcgarage/accelerated_dl_pytorch …Prikaži ovu nitHvala. Twitter će to iskoristiti za poboljšanje vaše vremenske crte. PoništiPoništi -
Marat Dukhan proslijedio/la je Tweet
The
#Spectre security flaw suddenly casts@OdedGreen's and@MaratDukhan's 2015 idea of avoiding branches altogether in a new light: https://dl.acm.org/citation.cfm?id=2755580 …pic.twitter.com/ClTF2p2d9b
Hvala. Twitter će to iskoristiti za poboljšanje vaše vremenske crte. PoništiPoništi -
#mInference demo, that won the 1st place on the AI Hackathon in Minsk, is now live on http://www.minference.com#WebAssembly#AsmJS#PNaClHvala. Twitter će to iskoristiti za poboljšanje vaše vremenske crte. PoništiPoništi -
Marat Dukhan proslijedio/la je Tweet
Andrew Tulloch and Yangqing Jia of Facebook give a shout out to
@MaratDukhan and NNPACK at#F8pic.twitter.com/Wc4XGiuCmc
Hvala. Twitter će to iskoristiti za poboljšanje vaše vremenske crte. PoništiPoništi -
Marat Dukhan proslijedio/la je Tweet
@MaratDukhan of@GTCSE demos NNPACK at the@MLatGT launch eventpic.twitter.com/4Cvj0fkA5y
Hvala. Twitter će to iskoristiti za poboljšanje vaše vremenske crte. PoništiPoništi -
A 1x1 layer computes a linear combination of per-channel images. But if the layer has 2x2 stride, it wastes 75% of computed pixels. (2/2)
Hvala. Twitter će to iskoristiti za poboljšanje vaše vremenske crte. PoništiPoništi -
Tonight I learnt: there are 1x1 convolutional layers with non-unit stride, e.g. in ResNet models. They make no sense, yet they exist. (1/2)
Hvala. Twitter će to iskoristiti za poboljšanje vaše vremenske crte. PoništiPoništi
Čini se da učitavanje traje već neko vrijeme.
Twitter je možda preopterećen ili ima kratkotrajnih poteškoća u radu. Pokušajte ponovno ili potražite dodatne informacije u odjeljku Status Twittera.

Blazeface is now available in browsers with TensorFlow.js!
The model detects faces and facial features in real-time 

Try it out now →
WASM has wider device support and better numerical stability while getting competitive with WebGL for smaller models.
Share your feedback here →