Xu Xing
Xu Xing
@qjia7 @xhcao @haoyunfeix @gyagp , This is for API discussion. The use case for this PR: ``` const shape: Shape = [16]; const gpuReadData = new tf.GPUReadData(gpuReadBuffer, shape, dtype); const...
One possible reason is, the vsc and clang-format are installed under different privilege or user. The simple way to prove this is: execute clang-format in vsc's terminal. If error reports,...
When debugging the handpose model, I found that (fused) conv2d may work not properly, with certain shape/stride/padding. With below two PRs applied, certain cases may works: For conv2d (non-fused): try...
It's great to see that. Thanks @NALLEIN. Two comments: 1), “used Conv2DMMProgram” ? Or used Conv2DNaiveProgram? Conv2DNaiveProgram should be the slow way. 2), “modifying the getAndSavePipelineh function” means to disable...
For the speed, after merge https://github.com/tensorflow/tfjs/issues/3095 and https://github.com/tensorflow/tfjs/pull/3049, I found that the handpose model runs faster. The fusedConv2D may worked correctly with https://github.com/tensorflow/tfjs/issues/3095, so if you mannually merge this PR,...
Sorry, a typo above, not 3095, 3095 is a issue, the fix of this issue is tensorflow/tfjs#3096
@qjia7 @xhcao @haoyunfeix @gyagp PTAL
@qjia7 @xhcao @haoyunfeix @gyagp PTAL
> I think it will impact the performance if you unconditionally add `pushErrorScope` and `popErrorScope`. Do you observe any perf loss? > > And in another side, I don't see...
@qjia7 @xhcao @haoyunfeix @gyagp PTAL