frozenleaves/pytorch
1
0
Fork 0
mirror of https://github.com/pytorch/pytorch.git synced 2025-10-20 21:14:14 +08:00
Code Issues Packages Projects Releases Wiki Activity

Updated OpInfos FAQ (markdown)

Matti Picus
2021-03-26 16:40:51 +03:00
parent 40c34fc405
commit 6659918351
1 changed files with 14 additions and 3 deletions
Download Patch File Download Diff File Expand all files Collapse all files

17
OpInfos-FAQ.md

@ -4,7 +4,7 @@ Here are some tips for when you want to write tests using OpInfos. The basic lay
```
class TestAdd(TestCase):
@ops([OpInfo1, OpInfo2]
@ops([OpInfo1, OpInfo2])
def test_ops(self, device, dtype, op):
for sample in op.sample_inputs(device, dtype):
tensors = sample.input
@ -18,12 +18,23 @@ instantiate_device_type_tests(TestAdd, globals())
<summary><b>How many tests will this run?</b></summary>
The `ops` decorator will create a set of tests for each `device` and `OpInfo`.
The `OpInfo.dtype*` and `OpInfo.default_test_dtypes` will be combined with the `device`
to further generate test functions.
to further generate test functions, each one will be called with a `device`, `dtype` and `op`.
</details>
<details>
<summary><b>How do I write a `sample_inputs` method?<b></summary>
<summary><b>The test signature has `device`, `dtype` and `op`, where do the rest of the parameters come from?</b></summary>
In order to generate tensors ([make_tensor](https://github.com/pytorch/pytorch/blob/61b074581ce1ccf0fb1bf4f1b73f4b99f93fa70c/torch/testing/_internal/common_utils.py#L1592)
is a convenient way to do that) you need some kind of **domain** of values ((high, low), non-negative, ...)
and shapes, as well as optional parameters like `requires_grad` and `contiguous`. These all can be part of the
OpInfo which has *Info* about the *Op* under test. The `sample_inputs` function is a convenient interface to
encapsulate this knowledge.
</details>
<details>
<summary><b>How do I write a `sample_inputs` method to generate tensors to test with?<b></summary>
TBD, include examples
</details>
<details>
<summary><b>A test failed. How do I reproduce it?<b></summary>
TBD