CUDA

I see comments suggesting adding this to understand how loops are being handled by numba, and in the their own FAQ (https://numba.pydata.org/numba-doc/latest/user/faq.html)

from llvmlite import binding as llvm
llvm.set_option('','--debug-only=loop-vectorize')

You would then create your njit function and run it, and I believe the idea is that it prints debug information about whether

It would be great to have FBeta, F2, or F0.5 metrics to be implemented without the need for a custom metric class defined by user.

catboost version: 0.26

the .pcd file format allows for fields to be extended. this means it can neatly hold data about the label or object of a point. this can be very handy for ML tasks. However, the open3d file io does not appear to be able to read other fields other than the xyz, rgb, normals etc . I haven't been able to find where in the open3d structure the code for the file io pcd loading is implemented to att

Is your feature request related to a problem? Please describe.

The row_equality_comparator constructor currently accepts a bool to indicate if nulls are considered equal or not. In libcudf, we prefer to use scoped enums instead of bools and we already have a [`

Current implementation of join can be improved by performing the operation in a single call to the backend kernel instead of multiple calls.

This is a fairly easy kernel and may be a good issue for someone getting to know CUDA/ArrayFire internals. Ping me if you want additional info.

First of all, great library!

I am having some confusion in understanding the role of values_first2 in set_difference_by_key as mentioned here.

In general terms, the result of set difference D = A - B will not contain values from B therefore values_first2 can never be part of D. So `val

请问可以直接training tmfile出来吗? 因为tengine-convert-tool covert 会有error

tengine-lite library version: 1.4-dev
Get input tensor failed

或是有例子能training出下面tmfile 呢?
![Screenshot from 2021-05-27 07-01-46](https://user-images.githubusercontent.com/40915044/11

Report needed documentation

Report needed documentation
While the estimator guide offers a great breakdown of how to use many of the tools in api_context_managers.py, it would be helpful to have information right in the docstring during development to more easily understand what is actually going on in each of the provided functions/classes/methods. This is particularly important for

In order to test manually altered IR, it would be nice to have a --skip-compilation flag for futhark test, just like we do for futhark bench.