[quant][graphmode][fx] Change standalone module api #49719
Closed
Conversation
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Summary:
We find there are multiple use cases for standalone module, one use case requires standalone module
to produce a module that takes float Tensor as input and outputs a float Tensor, the other needs to
produce a modulee that takes quantized Tensor as input and outputs a quantized Tensor.
This is similar to `quantized_input_idxs` and `quantized_output_idxs` so we want to nest
prepare_custom_config_dict in the standalone module configuration, for maximum flxibility we also
include qconfig_dict for stand alone module as well in case user needs to have special qconfig_dict for
the standalone module in the future.
Changed from
```python
prepare_custom_config_dict =
{
"standalone_module_name": ["standalone_module"],
"standalone_module_class": [StandaloneModule]
}
```
to
```python
prepare_custom_config_dict =
{
"standalone_module_name": [("standalone_module", qconfig_dict1, prepare_custom_config_dict1)],
"standalone_module_class": [(StandaloneModule, qconfig_dict2, prepare_custom_config_dict2)]
}
```
The entries in the config are:
1. name/module_class
2. optional qconfig_dict, when it is None, we'll use {"": qconfig} where qconfig is the one from parent qconfig_dict
3. optional prepare_custom_config_dict, when it is None, we'll use default value of prepare_custom_config_dict for prepare API (None)
Test Plan:
python test/test_quantization.py TestQuantizeFx.test_standalone_module
Reviewers:
Subscribers:
Tasks:
Tags:
[ghstack-poisoned]
💊 CI failures summary and remediationsAs of commit 07eab29 (more details on the Dr. CI page):
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 to the (internal) Dr. CI Users group. This comment has been revised 10 times. |
jerryzh168
added a commit
that referenced
this pull request
Dec 22, 2020
Summary:
We find there are multiple use cases for standalone module, one use case requires standalone module
to produce a module that takes float Tensor as input and outputs a float Tensor, the other needs to
produce a modulee that takes quantized Tensor as input and outputs a quantized Tensor.
This is similar to `quantized_input_idxs` and `quantized_output_idxs` so we want to nest
prepare_custom_config_dict in the standalone module configuration, for maximum flxibility we also
include qconfig_dict for stand alone module as well in case user needs to have special qconfig_dict for
the standalone module in the future.
Changed from
```python
prepare_custom_config_dict =
{
"standalone_module_name": ["standalone_module"],
"standalone_module_class": [StandaloneModule]
}
```
to
```python
prepare_custom_config_dict =
{
"standalone_module_name": [("standalone_module", qconfig_dict1, prepare_custom_config_dict1)],
"standalone_module_class": [(StandaloneModule, qconfig_dict2, prepare_custom_config_dict2)]
}
```
The entries in the config are:
1. name/module_class
2. optional qconfig_dict, when it is None, we'll use {"": qconfig} where qconfig is the one from parent qconfig_dict
3. optional prepare_custom_config_dict, when it is None, we'll use default value of prepare_custom_config_dict for prepare API (None)
Test Plan:
python test/test_quantization.py TestQuantizeFx.test_standalone_module
Reviewers:
Subscribers:
Tasks:
Tags:
ghstack-source-id: c5c100bd93256c96dc8dcdf9664a92d547e47403
Pull Request resolved: #49719
vkuzo
reviewed
Dec 22, 2020
torch/quantization/quantize_fx.py
Outdated
| @@ -178,11 +178,13 @@ def prepare_fx( | |||
| # optional: specify the path for standalone modules | |||
| # These modules are symbolically traced and quantized as one unit | |||
| "standalone_module_name": [ | |||
| "submodule.standalone" | |||
| # module_name, qconfig_dict, prepare_custom_config_dict | |||
| ("submodule.standalone", None, None) | |||
There was a problem hiding this comment.
does qconfig_dict == None here means "do not quantize" or "use parent qconfig"?
There was a problem hiding this comment.
it means "use parent qconfig"
for "do not quantize": qconfit_dict = {"": None}
| @@ -575,8 +575,8 @@ def forward(self, x): | |||
| original_ref_m.conv2.bias = torch.nn.Parameter(original_m.standalone.conv.bias.detach()) | |||
|
|
|||
| qconfig_dict = {"": default_qconfig} | |||
| config_name = {"standalone_module_name": ["standalone"]} | |||
| config_class = {"standalone_module_class": [StandaloneModule]} | |||
| config_name = {"standalone_module_name": [("standalone", None, None)]} | |||
There was a problem hiding this comment.
Can we also test for selective quantization of inputs/outputs for standalone modules?
There was a problem hiding this comment.
yes, it's in the next PR
There was a problem hiding this comment.
Sorry, just saw that
raghuramank100
approved these changes
Dec 22, 2020
Summary:
We find there are multiple use cases for standalone module, one use case requires standalone module
to produce a module that takes float Tensor as input and outputs a float Tensor, the other needs to
produce a modulee that takes quantized Tensor as input and outputs a quantized Tensor.
This is similar to `quantized_input_idxs` and `quantized_output_idxs` so we want to nest
prepare_custom_config_dict in the standalone module configuration, for maximum flxibility we also
include qconfig_dict for stand alone module as well in case user needs to have special qconfig_dict for
the standalone module in the future.
Changed from
```python
prepare_custom_config_dict =
{
"standalone_module_name": ["standalone_module"],
"standalone_module_class": [StandaloneModule]
}
```
to
```python
prepare_custom_config_dict =
{
"standalone_module_name": [("standalone_module", qconfig_dict1, prepare_custom_config_dict1)],
"standalone_module_class": [(StandaloneModule, qconfig_dict2, prepare_custom_config_dict2)]
}
```
The entries in the config are:
1. name/module_class
2. optional qconfig_dict, when it is None, we'll use {"": qconfig} where qconfig is the one from parent qconfig_dict
3. optional prepare_custom_config_dict, when it is None, we'll use default value of prepare_custom_config_dict for prepare API (None)
Test Plan:
python test/test_quantization.py TestQuantizeFx.test_standalone_module
Reviewers:
Subscribers:
Tasks:
Tags:
Differential Revision: [D25675704](https://our.internmc.facebook.com/intern/diff/D25675704)
[ghstack-poisoned]
Codecov Report
@@ Coverage Diff @@
## gh/jerryzh168/525/base #49719 +/- ##
==========================================================
- Coverage 80.67% 80.58% -0.09%
==========================================================
Files 1895 1895
Lines 205307 205316 +9
==========================================================
- Hits 165622 165446 -176
- Misses 39685 39870 +185 |
|
This pull request has been merged in f474ffa. |
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
Stack from ghstack:
Summary:
We find there are multiple use cases for standalone module, one use case requires standalone module
to produce a module that takes float Tensor as input and outputs a float Tensor, the other needs to
produce a modulee that takes quantized Tensor as input and outputs a quantized Tensor.
This is similar to
quantized_input_idxsandquantized_output_idxsso we want to nestprepare_custom_config_dict in the standalone module configuration, for maximum flxibility we also
include qconfig_dict for stand alone module as well in case user needs to have special qconfig_dict for
the standalone module in the future.
Changed from
to
The entries in the config are:
Test Plan:
python test/test_quantization.py TestQuantizeFx.test_standalone_module
Reviewers:
Subscribers:
Tasks:
Tags:
Differential Revision: D25675704