kunal-vaishnavi

Results 4 comments of kunal-vaishnavi

Issue with performing shape inference using symbolic_shape_infer.py with Phi-3 ONNX Models

The uploaded Phi-3 ONNX models already have been symbolic shape inferenced with dynamic axes. The symbolic shape inference for most quantization operators is defined in each operator's spec. https://github.com/microsoft/onnxruntime/blob/6baaaf516538f9059da3558b2cd22128a9e42c07/onnxruntime/core/graph/contrib_ops/contrib_defs.cc#L3434-L3481 Here...

Issue with performing shape inference using symbolic_shape_infer.py with Phi-3 ONNX Models

> The attached Phi-3 ONNX model is not shape inferred for all the operators. Couple of operators might have symbolic shape inferenced with dynamic axes. The vast majority of the...

fix: BART attention fusion for key with bias🐛

Thanks for your contribution! > I'd also like to add a test case. Could you provide some guidance where it should go? Add modelling code to `onnxruntime/test/python/transformers/test_bart.py`? Any feedback is...

[Transformers Optimizer] CLIP-ViT encoder attention not getting fused

Can you try your code with the [nightly ORT package](https://onnxruntime.ai/docs/get-started/with-python.html#builds) instead of the stable ORT 1.18.0 package? New fusions for CLIP were recently added in [this PR](https://github.com/microsoft/onnxruntime/pull/20721) and aren't in...