Mathieu Poumeyrol
Mathieu Poumeyrol
Yeah, it's slow, and not only because of Resize. You're in a model design space that is not my main focus: real time embedded voice application leads to relatively small...
No worries, did not mean to sound defensive on the slowness aspect :) just quickly writing down my findings... for you if you're curious or at least for my future...
Mmm... thanks for the heads up. No I don't think I have integrated yet to the test suite, and I definitely should. Let me get back to you on this...
Yeah, Resize is typically an operator for which we ourselves have very little interest, so we welcome contribution on it. I am aware of the limitations. I took an axis-per-axis...
No support right now, no drive to do them from our side, we could discuss how to add them as an optional extension to the curent TF operator set (in...
Mmmm... you do need an op for the STFT, indeed. But how do you foresee doing the binning ? Either... 0/ I'm missing something obvious 1/ you need a "general...
All right, let's see how the POC is going, I guess :)
`override-fact` is a dirty hack to override a fact in the case of inconsistent models (I have seen a couple where input and output are tagged with different batch size)....
Also if you run on actual input, the size should just induce the `N` value... Not sure what the --set brings to the table actually. Maybe for random inputs ?
Mmmm... I was thinking... you optimise the network with N, `-i input:N,3,224,224,f32 -O --allow-random-input` and then you want to profile it on random input for N=1 (or any other N...