pkgoogle

Hi @andrew-lyons, I believe you are correct that 15.0.1 does not allow 'force_load', though we are not xcode experts so you might want to check with them just in case....

Hi @christian-steinmeyer, the default behavior of the quantization system is to fallback to float32 ops if the converter finds it impossible (for various reasons) to convert the entire model to...

Hi @christian-steinmeyer, unfortunately, not that I currently know of .. there are some code changes coming up which will make it more obvious/known to end users (at least when it...

@christian-steinmeyer, Unfortunately not, I am not comfortable making a prediction yet on it's release date yet.

Hi @christian-steinmeyer, I don't see a reason why that shouldn't be fully convertable, @abattery, can you please take a look? Thanks.

Hi @Burton2000, apologies for all the whiplash but TF and TFL are a living codebases so everything is quite dynamic still. With 2.15 we are currently supporting dynamic input shapes:...

Hi @LakshmiKalaKadali, can you please take a look? Thanks.

Hi @jsonslim, this repo is intended to be used with https://developer.android.com/codelabs/digit-classifier-tflite. Have you tried specifically following all the steps in that tutorial? If you are still running into problems, then...

Hi @Black3rror, you can still run non-converted ops successfully and because they have higher precision the performance in terms of accuracy is actually usually better, however you lose the benefits...

Hi @Black3rror, if that is the configuration of your conversion then, you are correct, you should not expect any loss of precision, however with any quantization techniques included -- then...