debug ReduceSumDimInt

ReduceSumDimInt_debug.txt

https://github.com/llvm/torch-mlir/blob/6c1dea1c0ff22efb7119f6453655b8b38b52e506/lib/Conversion/TorchToLinalg/Reduction.cpp#L409
   ->
        result = convertScalarToDtype(rewriter, loc, result, mlir::IntegerType::get(op->getContext(), 64));


➜  SHARK git:(bloom) ✗ python tank/bloom_model.py
Some weights of BloomForSequenceClassification were not initialized from the model checkpoint at bigscience/bloom-560m and are newly initialized: ['score.weight']
You should probably TRAIN this model on a down-stream task to be able to use it for predictions and inference.
/home/chi/src/ubuntu20/shark/SHARK/shark.venv/lib/python3.10/site-packages/torch/jit/_check.py:181: UserWarning: The TorchScript type system doesn't support instance-level annotations on empty non-base types in `__init__`. Instead, either 1) use a type annotation in the class body, or 2) wrap the type in `torch.jit.Attribute`.
  warnings.warn("The TorchScript type system doesn't support "
/home/chi/src/ubuntu20/shark/SHARK/shark.venv/lib/python3.10/site-packages/torch/jit/_trace.py:744: UserWarning: The input to trace is already a ScriptModule, tracing it is a no-op. Returning the object as is.
  warnings.warn(
Traceback (most recent call last):
  File "/home/chi/src/ubuntu20/shark/SHARK/tank/bloom_model.py", line 95, in <module>
    module = torch_mlir.compile(
  File "/home/chi/src/ubuntu20/shark/torch-mlir/build/tools/torch-mlir/python_packages/torch_mlir/torch_mlir/__init__.py", line 273, in compile
    run_pipeline_with_repro_report(
  File "/home/chi/src/ubuntu20/shark/torch-mlir/build/tools/torch-mlir/python_packages/torch_mlir/torch_mlir/compiler_utils.py", line 73, in run_pipeline_with_repro_report
    raise TorchMlirCompilerError(trimmed_message) from None
torch_mlir.compiler_utils.TorchMlirCompilerError: Lowering Torch Backend IR -> Linalg-on-Tensors Backend IR failed with the following diagnostics:
error: 'linalg.yield' op type of yield operand 1 ('i64') doesn't match the element type of the enclosing linalg.generic op ('i1')
note: see current operation: "linalg.yield"(%31762) : (i64) -> ()


Error can be reproduced with:
$ torch-mlir-opt -pass-pipeline='torch-backend-to-linalg-on-tensors-backend-pipeline' /tmp/_lambda.mlir
Add '-mlir-print-ir-after-all -mlir-disable-threading' to get the IR dump for debugging purpose.





















https://github.com/llvm/torch-mlir/blob/6c1dea1c0ff22efb7119f6453655b8b38b52e506/lib/Conversion/TorchToLinalg/Reduction.cpp#L409
   ->
        result = convertScalarToDtype(rewriter, loc, result, mlir::IntegerType::get(op->getContext(), 1));
        

➜  SHARK git:(bloom) ✗ python tank/bloom_model.py
......
......
 x: array([128])
 y: array([ True])*; Dispatched function name: *aten.sum.dim_IntList*; Dispatched function args: *[torch.Size([1, 128]), [-1]]*; Dispatched function kwargs: *[]*; 
  warnings.warn(
Traceback (most recent call last):
  File "/home/chi/src/ubuntu20/shark/SHARK/shark/torch_mlir_lockstep_tensor.py", line 130, in __torch_dispatch__
    eager_module = build_mlir_module(func, normalized_kwargs)
  File "/home/chi/src/ubuntu20/shark/torch-mlir/build/tools/torch-mlir/python_packages/torch_mlir/torch_mlir/eager_mode/ir_building.py", line 343, in build_mlir_module
    assert len(annotations) == len(
AssertionError: Number of annotations and number of graph inputs differs.

During handling of the above exception, another exception occurred:

Traceback (most recent call last):
  File "/home/chi/src/ubuntu20/shark/SHARK/tank/bloom_model.py", line 37, in <module>
    output = model(eager_input_batch) # RuntimeError: Expected tensor for argument #1 'indices' to have one of the following scalar types: Long, Int; but got torch.FloatTensor instead (while checking arguments for embedding)
  File "/home/chi/src/ubuntu20/shark/SHARK/shark.venv/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1190, in _call_impl
    return forward_call(*input, **kwargs)
  File "/home/chi/src/ubuntu20/shark/SHARK/tank/bloom_model.py", line 23, in forward
    return self.model.forward(tokens)[0]
  File "/home/chi/src/ubuntu20/shark/SHARK/shark.venv/lib/python3.10/site-packages/transformers/models/bloom/modeling_bloom.py", line 955, in forward
    sequence_lengths = torch.ne(input_ids, self.config.pad_token_id).sum(-1) - 1
  File "/home/chi/src/ubuntu20/shark/SHARK/shark.venv/lib/python3.10/site-packages/torch/_tensor.py", line 1270, in __torch_function__
    ret = func(*args, **kwargs)
  File "/home/chi/src/ubuntu20/shark/SHARK/shark/torch_mlir_lockstep_tensor.py", line 198, in __torch_dispatch__
    out = func(*unwrapped_args, **unwrapped_kwargs)
  File "/home/chi/src/ubuntu20/shark/SHARK/shark.venv/lib/python3.10/site-packages/torch/_ops.py", line 60, in __call__
    return self._op(*args, **kwargs or {})
RuntimeError: Subtraction, the `-` operator, with a bool tensor is not supported. If you are trying to invert a mask, use the `~` or `logical_not()` operator instead.