In numba/stencils/stencil.py, there are various places (like line 552, "if isinstance(kernel_size[i][0], int):") where we check for "int" in relation to neighborhoods. I ran across a case where I was creating a neighborhood tuple by extracting values from a Numpy array. This causes a problem because those Numpy values will not match in these isinstance int checks. I worked around it by conver

Problem: the approximate method can still be slow for many trees
catboost version: master
Operating System: ubuntu 18.04
CPU: i9
GPU: RTX2080

Would be good to be able to specify how many trees to use for shapley. The model.predict and prediction_type versions allow this. lgbm/xgb allow this.

Describe the bug
Clipping a DataFrame or Series using ints causes a cudf Failure because it won't handle the different dtypes (int and float)

Steps/Code to reproduce bug

data = cudf.Series([-0.43, 0.1234, 1.5, -1.31])
data.clip(0, 1)

...
  File "cudf/_lib/replace.pyx", line 216, in cudf._lib.replace.clip
  File "cudf/_lib/replace.pyx", line 198, in cudf._lib.replace.clamp

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.

PR NVIDIA/cub#218 fixes this CUB's radix sort. We should:

• Check whether Thrust's other backends handle this case correctly.
• Provide a guarantee of this in the stable_sort documentation.
• Add regression tests to enforce this on all backends.

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

I often use -v just to see that something is going on, but a progress bar (enabled by default) would serve the same purpose and be more concise.

We can just factor out the code from futhark bench for this.