Add convert path for quantize_ QAT API #1540
Open
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: #1415 added a quantize_ QAT API for the prepare path. This commit adds the remaining convert path for users to actually perform end-to-end QAT using the quantize_ API. The new flow will look like: ``` from torchao.quantization import ( quantize_, int8_dynamic_activation_int4_weight, ) from torchao.quantization.qat import ( FakeQuantizeConfig, from_intx_quantization_aware_training, intx_quantization_aware_training, ) activation_config = FakeQuantizeConfig(torch.int8, "per_token", is_symmetric=False) weight_config = FakeQuantizeConfig(torch.int4, group_size=32) quantize_( my_model, intx_quantization_aware_training(activation_config, weight_config), ) quantize_(my_model, from_intx_quantization_aware_training()) quantize_(my_model, int8_dynamic_activation_int4_weight(group_size=32)) ``` Test Plan: python test/quantization/test_qat.py -k test_quantize_api_convert_path [ghstack-poisoned]
🔗 Helpful Links🧪 See artifacts and rendered test results at hud.pytorch.org/pr/pytorch/ao/1540
Note: Links to docs will display an error until the docs builds have been completed. ✅ No FailuresAs of commit e1327cd with merge base b5b739b ( This comment was automatically generated by Dr. CI and updates every 15 minutes. |
facebook-github-bot
added
the
CLA Signed
Summary: #1415 added a quantize_ QAT API for the prepare path. This commit adds the remaining convert path for users to actually perform end-to-end QAT using the quantize_ API. The new flow will look like: ``` from torchao.quantization import ( quantize_, int8_dynamic_activation_int4_weight, ) from torchao.quantization.qat import ( FakeQuantizeConfig, from_intx_quantization_aware_training, intx_quantization_aware_training, ) activation_config = FakeQuantizeConfig(torch.int8, "per_token", is_symmetric=False) weight_config = FakeQuantizeConfig(torch.int4, group_size=32) quantize_( my_model, intx_quantization_aware_training(activation_config, weight_config), ) quantize_(my_model, from_intx_quantization_aware_training()) quantize_(my_model, int8_dynamic_activation_int4_weight(group_size=32)) ``` Test Plan: python test/quantization/test_qat.py -k test_quantize_api_convert_path [ghstack-poisoned]
andrewor14
added
the
topic: improvement
Summary: #1415 added a quantize_ QAT API for the prepare path. This commit adds the remaining convert path for users to actually perform end-to-end QAT using the quantize_ API. The new flow will look like: ``` from torchao.quantization import ( quantize_, int8_dynamic_activation_int4_weight, ) from torchao.quantization.qat import ( FakeQuantizeConfig, from_intx_quantization_aware_training, intx_quantization_aware_training, ) activation_config = FakeQuantizeConfig(torch.int8, "per_token", is_symmetric=False) weight_config = FakeQuantizeConfig(torch.int4, group_size=32) quantize_( my_model, intx_quantization_aware_training(activation_config, weight_config), ) quantize_(my_model, from_intx_quantization_aware_training()) quantize_(my_model, int8_dynamic_activation_int4_weight(group_size=32)) ``` Test Plan: python test/quantization/test_qat.py -k test_quantize_api_convert_path [ghstack-poisoned]
Summary: #1415 added a quantize_ QAT API for the prepare path. This commit adds the remaining convert path for users to actually perform end-to-end QAT using the quantize_ API. The new flow will look like: ``` from torchao.quantization import ( quantize_, int8_dynamic_activation_int4_weight, ) from torchao.quantization.qat import ( FakeQuantizeConfig, from_intx_quantization_aware_training, intx_quantization_aware_training, ) activation_config = FakeQuantizeConfig(torch.int8, "per_token", is_symmetric=False) weight_config = FakeQuantizeConfig(torch.int4, group_size=32) quantize_( my_model, intx_quantization_aware_training(activation_config, weight_config), ) quantize_(my_model, from_intx_quantization_aware_training()) quantize_(my_model, int8_dynamic_activation_int4_weight(group_size=32)) ``` Test Plan: python test/quantization/test_qat.py -k test_quantize_api_convert_path [ghstack-poisoned]
jerryzh168
reviewed
Jan 10, 2025
Comment on lines
+1319
to
+1320
| quantize_(m, from_intx_quantization_aware_training()) | ||
| quantize_(m, int8_dynamic_activation_int4_weight(group_size=group_size)) |
There was a problem hiding this comment.
the scale/zero_point calculated from int8_dynamic_activation_int4_weight is not guaranteed to be the same as the ones from QAT right, is this OK?
There was a problem hiding this comment.
Yeah this is actually also the case in the old flow (Int8DynActInt4WeightQATQuantizer). I did verify that the qparams are the same today because we calculate them the same way before and after convert. If the user really wants to guarantee the same qparams they can also just store them somewhere and set them manually in the weight tensors, so I think it's OK
jerryzh168
approved these changes
Jan 10, 2025
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 (oldest at bottom):
Summary: #1415 added a quantize_
QAT API for the prepare path. This commit adds the remaining
convert path for users to actually perform end-to-end QAT using
the quantize_ API. The new flow will look like:
Test Plan:
python test/quantization/test_qat.py -k test_quantize_api_convert_path