You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
I wonder why the batch size is hard-coded to be 1 in the 'DatasetCalibrator' class.
There used to be an input argument int8_calib_batch_size in the 'torch2trt' function to specify the batch size for the calibration set. Is it now only possible to have batch size to be 1?
The text was updated successfully, but these errors were encountered:
Hi,
I wonder why the batch size is hard-coded to be 1 in the 'DatasetCalibrator' class.
There used to be an input argument
int8_calib_batch_size
in the 'torch2trt' function to specify the batch size for the calibration set. Is it now only possible to have batch size to be 1?The text was updated successfully, but these errors were encountered: