gpu

EDIT: The failure is due to update in Python 3.10 behaviour.

The following OpInfo tests fail locally (with Python 3.10) but pass on CI for gradient op

https://github.com/pytorch/pytorch/blob/97f29bda59deab8c063cf01f0a8ff4321b93c55e/torch/testing/_internal/common_methods_invocations.py#L8277-L8282

_________________________________

These APIs are deprecated a while ago, we'll want to get rid of them.

 λ ~/github/taichi master rg "@deprecated" python
python/taichi/lang/matrix.py
516:    @deprecated('ti.Matrix.transposed(a)', 'a.transpose()')
520:    @deprecated('a.T()', 'a.transpose()')
902:    @deprecated('ti.Matrix.var', 'ti.Matrix.field')
915:    @deprecated('ti.Vector.var', 'ti.Vector.field')
1142:    @depr

At this moment relu_layer op doesn't allow threshold configuration, and legacy RELU op allows that.
We should add configuration option to relu_layer.

usually, after trained model. i save model in cpp format with code:

cat_model.save_model('a', format="cpp")
cat_model.save_model('b', format="cpp")

but when my cpp need to use multi models.

in my main.cpp

#include "a.hpp"
#include "b.hpp"

int main() {
  // do something
  double a_pv = ApplyCatboostModel({1.2, 2.3});  // i want to a.hpp's model here
  double b_pv 

Hi ,

I have tried out both loss.backward() and model_engine.backward(loss) for my code. There are several subtle differences that I have observed , for one retain_graph = True does not work for model_engine.backward(loss) . This is creating a problem since buffers are not being retained every time I run the code for some reason.

Please look into this if you could.

Our users are often confused by the output from programs such as zip2john sometimes being very large (multi-gigabyte). Maybe we should identify and enhance these programs to output a message to stderr to explain to users that it's normal for the output to be very large - maybe always or maybe only when the output size is above a threshold (e.g., 1 million bytes?)

Implement GPU version of numpy.* functions in cupy.* namespace.

This is a tracker issue that lists the remaining numpy.* APIs (see also: comparison table). I've categorized them based on difficulty so that new contributors can pick the right task. Your contribution is highly welcomed and appreciated!

List of A

Is your feature request related to a problem? Please describe.
I have to place nvim\bin into path environment variable just so that neovide can sense nvim.exe.

Describe the solution you'd like
Add an option to specify nvim path in neovide settings.

https://github.com/brendangregg/FlameGraph

Describe the bug

W-THU is not supported right now. It only works if user specifies seasonal_period in setup

To Reproduce

Expected behavior
Would have expected it to automatically infer the freq = 7 from there.

For pandas API compatibility, we can implement Series.autocorr. autocorr calculates the Pearson correlation between the Series and itself lagged by N steps. Conceptually, this is a combination of shift and corr.

import pandas as pd
​
s = pd.Series([0.25, 0.5, 0.2, -0.05])
print(s.autocorr())
print(s

环境

1.系统环境：

• Ubuntu 20.04.2 LTS x86_64
• GLIBCXX: 3.4.28 (libstdc++.so.6.0.28)
• GPU:
  

2.MegEngine版本：1.6.0rc1
3.python版本：Python 3.8.10

The program stuck at net.load when I was trying to use the MegFlow. I wait for more than 10min and there is no sign of finishing it.

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.