Support quantized and non-quantized HF briaai/RMBG-1.4 conversion - #3
Open
mtavenrath wants to merge 2 commits into
Open
Support quantized and non-quantized HF briaai/RMBG-1.4 conversion#3mtavenrath wants to merge 2 commits into
mtavenrath wants to merge 2 commits into
Conversation
Handle zero-element optional tensors, fold Cast/Slice shape subgraphs, and infer Resize, Expand, Tile, Squeeze, Split, Range, and normalization output shapes. This fixes WebNN conversion of Hugging Face briaai/RMBG-1.4 non-quantized FP32 and FP16 ONNX exports.
Add DynamicQuantizeLinear and ConvInteger lowering through WebNN primitives, including zero-point handling and output type inference. Improve shape propagation, constant folding, Resize size handling, and packed quantized tensor decoding to successfully convert briaai/RMBG-1.4 FP32, FP16, and quantized ONNX exports. Add regression coverage, update operator support documentation, and remove model-specific shape-inference debug probes.
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
Add DynamicQuantizeLinear and ConvInteger lowering through WebNN
primitives, including zero-point handling and output type inference.
Improve shape propagation by handling zero-element optional tensors,
folding Cast/Slice shape subgraphs, and inferring output shapes for
Resize, Expand, Tile, Squeeze, Split, Range, and normalization ops.
Enhance constant folding, Resize size handling, and packed quantized
tensor decoding.
Add regression coverage, update operator support documentation, and
remove model-specific shape-inference debug probes.