Luca Antiga

Wouldn't it make sense to turn it around and have an `enable-performance` flag instead?

My need is about the now though (while designing the training app): how can we mount a lit:// drive? Should we surface its underlying s3 bucket? Here's the use case:...

Any feedback on the strategy to enable mounting `lit://` drives RO from an API perspective? A cheap way (in a future PR) would be in this case to snapshot a...

Thank you @tqchen, there seems to be different ways in which the different frameworks decide to implement bool. It is not guaranteed to be `uint(1)`, for instance in PyTorch is...

Hi @Akhtar303, I'm not sure the issue is with AVX. Stock libtensorflow that we download for x86 is not compiled with AVX instructions AFAIK (it's the one provided by TensorFlow,...

Yes, you can build libtensorflow from scratch on your machine. Take a look at the Makefile and Dockerfile for x86, this way we'll be able to rule out that it's...

Hi @Akhtar303, it looks you are triggering the arm32 build. I suggest you build for x64 on your machine using the Dockerfile we provide: https://github.com/RedisAI/RedisAI/blob/master/opt/build/tensorflow/Dockerfile.x64 You should be able to...

Hi, sorry for not replying until now. Can you help us identify the page where the mistake is located?

Hi @bkowshik, thanks for the issue. The error is saying that `target` is a Float tensor, while it should be Long, since it's been used as an index. Towards the...

What do we think about `connect` being env var based by default (a la conda)? I’ve been bitten by forgetting to disconnect multiple times, assuming that closing the terminal session...