Guillermo Gabrielli

You need to use a CUDA capable GPU i.e. Nvidia to train anything larger than a toy dataset in a reasonable time, that's how computer vision works in practice. You...

Is that 640x640, 80 classes? This is just a guess from what I remember from YoloV3. I think that's 1 image x 3 scales (a 32x32 px grid, a 16x16...

When preserve_alignments = True and compute_timestamps = True is true on a RNNT model, it looks like the timestep in the hypothesis becomes a dict, and self.timestep on [rnnt_utils.py:117](https://github.com/NVIDIA/NeMo/blob/d7658b57b196742ec6f517963a20e3a4de3983c2/nemo/collections/asr/parts/utils/rnnt_utils.py#L117) has...

YOLOv9 = GELAN + PGI, GELAN weights have no PGI (see section 5.5 of the paper). They have indeed slighly different results. Converted-c weights results are identical to the original...

Try to downgrade Pillow pip install "Pillow