JTan

That's strange - are you are training from scratch completely? If so, you may need to train the base model (warmup phase) for a longer period than you are currently,...

Yeah I think somewhere along the line I modified the models to make them more efficient and this got out of sync with the pretrained models - so you'll have...

For image regions with high-frequency detail, e.g. faces, text, this model tends not to do well, presumably because OpenImages does not contain many examples of these images, or perhaps because...

Posting the full stacktrace would help. If you rename the dataset in `default_config.py` under `DatasetPaths` you must also create a new dataset with corresponding name which inherits from the `BaseDataset`...

If you post the stacktrace it would be easier to diagnose. If you look at the parent `BaseDataset` class you'll notice the dataset directory should contain `train/` and `test/` subfolders.

I think the problem was the following line: ```python self.imgs = glob.glob(os.path.join(data_dir, '*.jpg')) ``` which would only get JPGs. I pushed a fix to master to account for PNGs as...

Yes, images are encoded in `.hific` format - to decode this back to an image you need to first: 1. Decode the bitstream to the latent representation using a standard...

Hi all, Sorry for the slow response. I think the theoretical bpp is computed as the entropy of the latents and hyperlatents as output by their respective probability models. On...

Observing similar behaviour with `jax.lax.scan` when trying to implement the logarithm of the cumulative sum of exponentials using a prefix scan. Here's a MWI: ```python import jax import jax.numpy as...

Just ran into this error - could you explain briefly why this error occurs? I'd expect the call to `__init__` to invoke the `setup` method. In the docs (https://flax.readthedocs.io/en/latest/flax.linen.html#flax.linen.Module.setup), it...