Mark Daoust

@EnggQasim we can't move forward with this if you don't sign the CLA.

Hi Karl, the TensorFlow site has fewer people/resources devoted to it now. The old system where we were in the loop for doc updates probably isn't feasible any more. What...

Right now it's different each time. They're working on it.

Yeah, I was hitting the same problem trying to speed up [this tutorial](https://www.tensorflow.org/tutorials/video/video_classification) by converting it to tfio. But you can't use `from_ffmpeg` in a dataset.map because it's graph mode...

> what indentation changes are causing this issue? I didn't indent anything like the original file so I'm not sure The diff is a lot easier to review if you...

> Empty notebook Good idea, we can point it to here: https://cloud.google.com/vertex-ai/generative-ai/docs/models/gemini-use-supervised-tuning#google-gen-ai-sdk

@httplups The object it returns is unordered. It may just be printed alphabetically, the order it generates them may just be incidental. > propertyOrdering The API could add something like...

@IsaSalemHame the API has added a propertyOrdering field. But we are slowly reducing development on this SDK. propertyOrdering is supported in the new SDK we released with Gemini-2: `google-genai` -...

Hi, thanks for the report. There have been cases where bad inputs cause 500s. But it sounds like it wasn't consistent about which requests cause the error. Like @gmKeshari said...

A 500 error can be generated for a variety of reasons. It feels like this issue isn't really connected to the inspect framework, and is just collecting any reports of...