funasr/models/ctc/ctc.py
@@ -40,9 +40,7 @@ logging.warning("ignore_nan_grad option is not supported for warp_ctc") self.ctc_loss = warp_ctc.CTCLoss(size_average=True, reduce=reduce) else: raise ValueError( f'ctc_type must be "builtin" or "warpctc": {self.ctc_type}' ) raise ValueError(f'ctc_type must be "builtin" or "warpctc": {self.ctc_type}') self.reduce = reduce
