Zoobot

Zoobot now includes pretrained greyscale encoders. FinetuneableZoobot... classes can use these pretrained encoders.

This [HuggingFace collection](https://huggingface.co/collections/mwalmsley/zoobot-encoders-greyscale-66427c51133285ca01b490c6) lists all current greyscale encoders.

What's Changed

- New pretrained architectures: ConvNeXT, EfficientNetV2, MaxViT, and more. Each in several sizes. Available on HuggingFace.
- Reworked finetuning procedure. All these architectures are finetuneable through a common method.
- Reworked finetuning options. Batch norm finetuning removed. Cosine schedule option added.
- Reworked finetuning saving/loading. Auto-downloads encoder from HuggingFace.
- Now supports regression finetuning (as well as multi-class and binary). See `pytorch/examples/finetuning`
- Updated `timm` to 0.9.10, allowing latest model architectures. Previously downloaded checkpoints may not load correctly!
- (internal until published) GZ Evo v2 now includes Cosmic Dawn (HSC H2O). Significant performance improvement on HSC finetuning. Also now includes GZ UKIDSS (dragged from our archives).
- Updated `pytorch` to `2.1.0`
- Added support for webdatasets (only recommended for large-scale distributed training)
- Improved per-question logging when training from scratch
- Added option to compile encoder for max speed (not recommended for finetuning, only for pretraining).
- Deprecates TensorFlow. The CS research community focuses on PyTorch and new frameworks like JAX.

**Full Changelog**: https://github.com/mwalmsley/zoobot/compare/v1.0.5...v2.0

What's Changed

Major improvements to finetuning.
- Now supports finetuning resnet and maxvit-tiny! This previously only worked as static pretrained models.
- Now specify num_blocks instead of num_layers, reflecting the network structure (effnet, resnet50, maxvit)
- Now you can optionally keep batchnorm layers always trainable

Small changes:
Update dependency to use galaxy-datasets==0.15.0
Replace multi-class example with new galaxy-mnist dataset
Fixes https://github.com/mwalmsley/zoobot/issues/108

The next release will update timm to 0.9ish
Pytorch 2.1.0 already works, and the next release will also reflect that

**Full Changelog**: https://github.com/mwalmsley/zoobot/compare/v1.0.4...v1.0.5

This minor release adds pretrained *color* models (i.e. with 3 channels) and also adds support for multi-class finetuning.

What's Changed
* sync by mwalmsley in https://github.com/mwalmsley/zoobot/pull/92
* 1.0.3 version bump by mwalmsley in https://github.com/mwalmsley/zoobot/pull/99
* Rebuild docs by mwalmsley in https://github.com/mwalmsley/zoobot/pull/103
* Add color weights, trivial QoL changes by mwalmsley in https://github.com/mwalmsley/zoobot/pull/104
* Add multiclass example by mwalmsley in https://github.com/mwalmsley/zoobot/pull/105
* Add multiclass support by mwalmsley in https://github.com/mwalmsley/zoobot/pull/106


**Full Changelog**: https://github.com/mwalmsley/zoobot/compare/v1.0.3...v1.0.4

Minor release marking Zoobot's state on acceptance of the JOSS paper.

What's Changed
* Finetune v1 onto docs by mwalmsley in https://github.com/mwalmsley/zoobot/pull/85
* Dev by mwalmsley in https://github.com/mwalmsley/zoobot/pull/87
* sync by mwalmsley in https://github.com/mwalmsley/zoobot/pull/93
* Delete requirements.txt by mwalmsley in https://github.com/mwalmsley/zoobot/pull/96
* Bring paper fix from docs to main by mwalmsley in https://github.com/mwalmsley/zoobot/pull/97
* Highlight Colab notebook by mwalmsley in https://github.com/mwalmsley/zoobot/pull/98


**Full Changelog**: https://github.com/mwalmsley/zoobot/compare/1.0.2...v1.0.3

- Lightning v2 now correctly uses early stopping
- galaxy-datasets version bumped to correctly ignore alpha channel of png images

What's Changed
* Hotfix - add early stopping back, update docs by mwalmsley in https://github.com/mwalmsley/zoobot/pull/91


**Full Changelog**: https://github.com/mwalmsley/zoobot/compare/v1.0.1...1.0.2