[quant] torch.max_pool1d #45152
[quant] torch.max_pool1d #45152
Conversation
[ghstack-poisoned]
💊 CI failures summary and remediationsAs of commit 045cb68 (more details on the Dr. CI page): 💚 💚 Looks good so far! There are no failures yet. 💚 💚 This comment was automatically generated by Dr. CI (expand for details).Follow this link to opt-out of these comments for your Pull Requests.Please report bugs/suggestions on the GitHub issue tracker or post in the (internal) Dr. CI Users group. This comment has been revised 22 times. |
Differential Revision: [D23846473](https://our.internmc.facebook.com/intern/diff/D23846473) [ghstack-poisoned]
| // (C, L) -> (C, 1, L) => kSqueezeDim = 1 | ||
| // (N, C, L) -> (N, C, 1, L) => kSqueezeDim = 2 |
There was a problem hiding this comment.
just double checking, this works as expected across different memory formats, right?
There was a problem hiding this comment.
Can you elaborate -- are you asking if squeeze/unsqueeze works properly across the memory formats, or if this trick does? For the latter -- it does, that what they do for the FP pooling ops. As for the squeeze/unsqueeze -- not sure, need to check the ref
There was a problem hiding this comment.
yes, was just asking if this function will work as expected for various memory formats. If it does - sg.
Codecov Report
@@ Coverage Diff @@
## gh/z-a-f/66/base #45152 +/- ##
=================================================
Coverage 68.05% 68.05%
=================================================
Files 396 396
Lines 51232 51238 +6
=================================================
+ Hits 34865 34870 +5
- Misses 16367 16368 +1
Continue to review full report at Codecov.
|
Differential Revision: [D23846473](https://our.internmc.facebook.com/intern/diff/D23846473) [ghstack-poisoned]
Differential Revision: [D23846473](https://our.internmc.facebook.com/intern/diff/D23846473) [ghstack-poisoned]
| _packed_params = torch.ops.quantized.linear_prepack(weight, bias) | ||
| return torch.ops.quantized.linear(input, _packed_params, scale, zero_point) | ||
|
|
||
| def max_pool1d(input, kernel_size, stride=None, padding=0, dilation=1, |
There was a problem hiding this comment.
do we need this? i feel user won't be directly using these things in general
There was a problem hiding this comment.
This is to support the similarity to the nn.functional. Should I remove it?
Differential Revision: [D23846473](https://our.internmc.facebook.com/intern/diff/D23846473) [ghstack-poisoned]
Stack from ghstack:
Differential Revision: D23846473