mit-han-lab / tinyengine

[NeurIPS 2020] MCUNet: Tiny Deep Learning on IoT Devices; [NeurIPS 2021] MCUNetV2: Memory-Efficient Patch-based Inference for Tiny Deep Learning; [NeurIPS 2022] MCUNetV3: On-Device Training Under 256KB Memory
https://mcunet.mit.edu
MIT License
806 stars 131 forks source link

Code Generation Patch Based Inference Bug #27

Closed Zerkclown closed 1 year ago

Zerkclown commented 1 year ago

Facing an issue related to PR #26:

Traceback (most recent call last): File "examples/vww.py", line 31, in life_cycle_path="./lifecycle.png", File "/Users/amahmed/Desktop/UMass/Spring_2022/Thesis/tinyengine/code_generator/CodegenUtilTFlite.py", line 70, in GenerateSourceFilesFromTFlite code_generator.codeGeneration() File "/Users/amahmed/Desktop/UMass/Spring_2022/Thesis/tinyengine/code_generator/CodeGenerator.py", line 131, in codeGeneration self._genPatchInference() File "/Users/amahmed/Desktop/UMass/Spring_2022/Thesis/tinyengine/code_generator/CodeGenerator.py", line 182, in _genPatchInference last_patch_op_output_buffer_str_for_patch_inference = last_patch_op._getBufferstr( AttributeError: 'NoneType' object has no attribute '_getBufferstr'

Deleting lines 179 to 181 in CodeGenerator.py removes the error.

meenchen commented 1 year ago

Hi @Zerkclown,

Thanks for reporting the bug. The fix has been merged: https://github.com/mit-han-lab/tinyengine/pull/28.

Zerkclown commented 1 year ago

Thank you, this fixed it!