ry3s opened a new issue, #15931: URL: https://github.com/apache/tvm/issues/15931
Thanks for participating in the TVM community! We use https://discuss.tvm.ai for any general usage questions and discussions. The issue tracker is used for actionable items such as feature proposals discussion, roadmaps, and bug tracking. You are always welcomed to post on the forum first :smile_cat: Issues that are inactive for a period of time may get closed. We adopt this policy so that we won't lose track of actionable issues that may fall at the bottom of the pile. Feel free to reopen a new one if you feel there is an additional problem that needs attention when an old one gets closed. ### Expected behavior Succeed in `FakeQuantizationToInteger`. ### Actual behavior Segumentation fault. ### Environment * TVM: [fa4aeee64efbd55db1f502a220276f9851c52f15](https://github.com/apache/tvm/tree/fa4aeee64efbd55db1f502a220276f9851c52f15) * onnx: 1.13.1 * onnxruntime-gpu: 1.14.1 * torch: 2.1.0 ### Steps to reproduce ```python import numpy as np import onnx import torch from onnxruntime.quantization import CalibrationDataReader, quantize_static, QuantFormat from torch import nn import tvm from tvm import relay INPUT_SHAPE = (1, 24, 24, 3) def write_relay_to_txt(obj, filename): with open(filename, "w") as f: f.write(relay.astext(obj, show_meta_data=False)) class RandomDataReader(CalibrationDataReader): def __init__(self) -> None: super().__init__() self.i = 0 def get_next(self) -> dict | None: if self.i >= 100: return None self.i += 1 return {"input": np.random.rand(*INPUT_SHAPE).astype(np.float32)} def main(): model = nn.Linear(3, 3, bias=False) input = torch.rand(INPUT_SHAPE, dtype=torch.float32) torch.onnx.export(model, input, "linear.onnx", input_names=["input"], output_names=["output"]) reader = RandomDataReader() quantize_static( "linear.onnx", "linear_quant.onnx", reader, per_channel=True, quant_format=QuantFormat.QDQ ) onnx_model = onnx.load("linear_quant.onnx") mod, param = relay.frontend.from_onnx(onnx_model, shape={"input": INPUT_SHAPE}) write_relay_to_txt(mod, "onnx_relay.txt") with tvm.transform.PassContext(opt_level=2): mod = relay.transform.InferType()(mod) mod = relay.transform.SimplifyInference()(mod) mod = relay.transform.FakeQuantizationToInteger()(mod) write_relay_to_txt(mod, "relay.txt") lib = relay.build(mod, target="llvm", params=param) lib.export_library("linear_quant.tar") if __name__ == "__main__": main() ``` ### Triage Please refer to the list of label tags [here](https://github.com/apache/tvm/wiki/Issue-Triage-Labels) to find the relevant tags and add them below in a bullet format (example below). * needs-triage -- This is an automated message from the Apache Git Service. To respond to the message, please log on to GitHub and use the URL above to go to the specific comment. To unsubscribe, e-mail: [email protected] For queries about this service, please contact Infrastructure at: [email protected]
