fix(examples): QuantVGG classifier in_features after TruncAvgPool2d (#1500) - #1630
Open
tonycoder-hub wants to merge 1 commit into
Open
tonycoder-hub wants to merge 1 commit into
tonycoder-hub wants to merge 1 commit into
Conversation
…ilinx#1500) TruncAvgPool2d(kernel_size=7, stride=1) on the 7x7 feature map from VGG's five MaxPools yields a 1x1 spatial map, so flatten size is 512 channels, not 512*7*7 (that value matches torchvision's AdaptiveAvgPool2d((7,7))). Signed-off-by: Tony Coder <407243179@qq.com>
This branch has not been deployed
This file contains hidden or 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
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
1 participant
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
Summary
Fixes #1500.
QuantVGGusesTruncAvgPool2d(kernel_size=(7, 7), stride=1)after VGG's fiveMaxPool2dlayers (224→7 spatial). That pool collapses 7×7 to 1×1, sotorch.flatten(x, 1)has size 512, not512 * 7 * 7.The previous
QuantLinear(512 * 7 * 7, 4096, …)matched torchvision VGG, which keeps 7×7 viaAdaptiveAvgPool2d((7, 7)). WithTruncAvgPool2dit raises a shape mismatch on any standard 224×224 input.Changes
QuantLinearin_featuresto512.tests/brevitas_examples/test_quant_vgg_shape.py— forwardquant_vgg11on(2, 3, 224, 224)and assert(2, 1000).Test plan
pytest tests/brevitas_examples/test_quant_vgg_shape.py -qquant_vgg11()(torch.randn(1,3,224,224)).shape == (1,1000)Signed-off-by: Tony Coder 407243179@qq.com