Deepak Narayanan

Hi @hustnn, The conversion is happening at runtime, but I do agree that it would be nice to do the conversion directly over the native column type (in this case,...

Hi @wjliu, Can you maybe try calling [this free function](https://github.com/weld-project/weld/blob/master/python/weld/bindings.py#L96) to see if it helps (you might need to wrap `result_list` in a [`WeldValue` object](https://github.com/weld-project/weld/blob/master/python/weld/bindings.py#L73) first though)? Let me know...

Hmm, I might need to do some work here to get this to work cleanly. (basically we want the `WeldValue` returned by the Weld program to be stored somewhere in...

Yeah, this makes sense. I don't think we're wed to any particular data format at this point -- it would be nice if Weld uses the same data format Arrow...

Hi @hustnn, Thanks for your interest in the project! I have Pandas 0.19.2 on my machine as well. I'm seeing these times (I might be using a different dataset to...

Did you use a PyTorch NGC container? If so, what version?

What's the use case for releasing the model memory? Trying to delete the optimizer object might help with releasing the optimizer memory (so something like `del optimizer` in `megatron/training.py`).

I see, makes sense. You probably don't want to release model memory still. Presumably you are not doing a backward pass the first time you run the training function? You...

Hmm, not sure. `del model` and `del optimizer` seem like the "right" things to do here.

Thanks for this, @jamesr66a! It is quite detailed; it will take me a few more days to give it the full attention it deserves. Some very quick high-level thoughts: -...