-
i was trying to follow notebooks with yolo v8 and v7. Traceback (most recent call last): How can I fix this error? I suppose I need to apply ignore_scope with: Here is my code.
|
Beta Was this translation helpful? Give feedback.
Replies: 1 comment 4 replies
-
Hi @MinGiSa, NNCF does not support quantization of custom PyTorch modules with weights. For example, "yolo_nas_l" has Taking into account that this module is the head of the model I would recommend to ignore it to preserve accuracy of the quantized model:
BTW: We are working on the fix of this issue #2461. Stay tune! |
Beta Was this translation helpful? Give feedback.
Hi @MinGiSa,
NNCF does not support quantization of custom PyTorch modules with weights. For example, "yolo_nas_l" has
NDFLHeads
module which callsconv2d
function withself.proj_conv
in forward which can not be quantized automaticly: https://github.com/Deci-AI/super-gradients/blob/7067736cb9062245aa4f118d91b03bf8de898ef7/src/super_gradients/training/models/detection_models/yolo_nas/dfl_heads.py#L210.Taking into account that this module is the head of the model I would recommend to ignore it to preserve accuracy of the quantized model: