mariecwhite
mariecwhite
Many TFLite models are failing on vulkan with the error `failed to legalize operation 'arith.constant'`: ``` error: failed to legalize operation 'arith.constant' note: called from error: failed to run translation...
### What happened? When importing [fsaf-cppe5_1.tflite](https://tfhub.dev/rishit-dagli/fsaf-cppe5/1), getting the error: ``` fsaf-cppe5_1.tflite:0:0: error: 'tosa.gather' op operand #1 must be 2D tensor of 32-bit signless integer values, but got 'tensor' fsaf-cppe5_1.tflite:0:0: note:...
For each benchmark run, please include in the results: - details about tuning config, compiler flags used, etc. - details about runtime parameters e.g. number of threads used - any...
We've identified some models that we would like to track closely. Some of these are not in the Shark Tank. Can you please add the following: - [ ] **GPT2...
### What happened? When compiling ResNet-V1-101 with dynamic shapes, getting the error: ``` resnet_v1_101_dynamic.mlir:223:12: error: failed to legalize operation 'tosa.max_pool2d' %220 = "tosa.max_pool2d"(%219) {kernel = [3, 3], pad = [1,...
For MobileNetV2 UINT8, peak resident set size on IREE is 2-3x larger on both mobile and server CPU compared with TFLite. For all other models under test, IREE memory use...
ONNX is SOTA for a lot of models that we are benchmarking. Can this be included in the benchmarks?
For GPU, it would be great to benchmark models in both FP16 and FP32. In IREE/SHARK, this is a matter of changing compiler flags but not so clear on how...
Please add RNNT (speech recognition) to the Shark Tank: https://github.com/mlcommons/inference/tree/master/speech_recognition/rnnt
TorchInductor is outperforming other PyTorch compilers so it would be great if we can use this as a baseline. Reference: https://dev-discuss.pytorch.org/t/torchinductor-a-pytorch-native-compiler-with-define-by-run-ir-and-symbolic-shapes/747