whisper-base pytorch model support

[AmosLewis]

iree-compiler 20240620.930
iree-runtime 20240620.930

python runmodel.py  --torchmlirimport fximport --todtype default --mode onnx --outfileprefix whisper-base 1> model-run.log 2>&1
iree-import-onnx whisper-base.default.onnx -o whisper-base.default.pytorch.linalg.mlir 1> onnx-import.log 2>&1
iree-compile --iree-input-demote-i64-to-i32 --iree-hal-target-backends=llvm-cpu  whisper-base.default.pytorch.linalg.mlir > whisper-base.default.vmfb 2>iree-compile.log

whisper-base.default.pytorch.linalg.mlir:139:12: error: 'linalg.generic' op inferred input/output operand #1 has shape's dimension #1 to be 8, but found 1
    %135 = torch.operator "onnx.Add"(%134, %106) : (!torch.vtensor<[1,8,10,10],f32>, !torch.vtensor<[?,?,10,10],f32>) -> !torch.vtensor<[?,8,10,10],f32> 
           ^
whisper-base.default.pytorch.linalg.mlir:139:12: note: see current operation: 
%397 = "linalg.generic"(%395, %318, %396) <{indexing_maps = [affine_map<(d0, d1, d2, d3) -> (0, d1, d2, d3)>, affine_map<(d0, d1, d2, d3) -> (d0, d1, d2, d3)>, affine_map<(d0, d1, d2, d3) -> (d0, d1, d2, d3)>], iterator_types = [#linalg.iterator_type<parallel>, #linalg.iterator_type<parallel>, #linalg.iterator_type<parallel>, #linalg.iterator_type<parallel>], operandSegmentSizes = array<i32: 2, 1>}> ({
^bb0(%arg1182: f32, %arg1183: f32, %arg1184: f32):
  %1727 = "arith.addf"(%arg1182, %arg1183) <{fastmath = #arith.fastmath<none>}> : (f32, f32) -> f32
  "linalg.yield"(%1727) : (f32) -> ()
}) : (tensor<1x8x10x10xf32>, tensor<1x1x10x10xf32>, tensor<1x8x10x10xf32>) -> tensor<1x8x10x10xf32>

python ./run.py --tolerance 0.001 0.001 --cachedir ./huggingface_cache  -f pytorch -g models --mode onnx --report  --torchtolinalg -j 12
Starting e2eshark tests. Using 12 processes
Cache Directory: /home/chi/src/SHARK-TestSuite/e2eshark/huggingface_cache
Tolerance for comparing floating point (atol, rtol) = (0.001, 0.001)
Note: No Torch MLIR build provided using --torchmlirbuild. iree-import-onnx will be used to convert onnx to torch onnx mlir
IREE BUILD: IREE in PATH /home/chi/src/SHARK-TestSuite/e2eshark/e2e_venv/bin will be used
Test run directory: /home/chi/src/SHARK-TestSuite/e2eshark/test-run
Framework:pytorch mode=onnx backend=llvm-cpu runfrom=model-run runupto=inference
Test list: ['pytorch/models/gpt2', 'pytorch/models/opt-125M', 'pytorch/models/bart-large', 'pytorch/models/whisper-base', 'pytorch/models/beit-base-patch16-224-pt22k-ft22k', 'pytorch/models/deit-small-distilled-patch16-224', 'pytorch/models/resnet50', 'pytorch/models/llama2-7b-GPTQ', 'pytorch/models/mobilebert-uncased', 'pytorch/models/vicuna-13b-v1.3', 'pytorch/models/phi-1_5', 'pytorch/models/opt-350m', 'pytorch/models/vit-base-patch16-224', 'pytorch/models/phi-2', 'pytorch/models/dlrm', 'pytorch/models/gpt2-xl', 'pytorch/models/opt-125m-gptq', 'pytorch/models/whisper-medium', 'pytorch/models/stablelm-3b-4e1t', 'pytorch/models/opt-1.3b', 'pytorch/models/whisper-small', 'pytorch/models/t5-large', 'pytorch/models/gemma-7b', 'pytorch/models/miniLM-L12-H384-uncased', 'pytorch/models/mit-b0', 'pytorch/models/bert-large-uncased', 'pytorch/models/t5-base', 'pytorch/models/llama2-7b-hf', 'pytorch/models/bge-base-en-v1.5']
Test pytorch/models/resnet50 passed
Test pytorch/models/deit-small-distilled-patch16-224 passed
Test pytorch/models/whisper-base failed [iree-compile]