Torchcodec has a mechanism to report CPU fallback. See: * https://github.com/meta-pytorch/torchcodec/blob/d1a1197d985804f55f77406a176abc796960591b/test/test_decoders.py#L2003 * https://github.com/meta-pytorch/torchcodec/blob/d1a1197d985804f55f77406a176abc796960591b/src/torchcodec/decoders/_video_decoder.py#L26 We also might need to generalize approach in Torchcodec before fully implementing this. CC: @eromomon
Torchcodec has a mechanism to report CPU fallback. See:
We also might need to generalize approach in Torchcodec before fully implementing this.
CC: @eromomon