[WIP] make repr formatting narrower

[davanstrien]

What this PR does?

This PR aims to make the __repr__s in hf_api.py easier to view on a narrow screen i.e. to favour a longer __repr__ for a class over a short but very wide __repr__. This PR closes #857

Intended outcome

As discussed in #857, the goal here is to have a __repr__ that is easy to view on a narrow screen without horizontal scrolling.

I have tried to make sure the approach taken doesn't do anything esoteric and doesn't add too much additional code to the __repr__. I have tried to keep the format of the __repr__ as similar to what is there now, except for making it narrower.

Approach taken

For now, I have just changed the __repr__ for ModelInfo to get initial feedback. The __repr__ now uses pprint.pformat to offload the formatting work for the attributes inside the class. At the moment, the __repr__s for all the classes in hf_api.py show all the attributes, but it would also be possible to hide some before passing to pformat if needed. I chose this approach since many people working with python will be at least vaguely familiar with pprint so hopefully, the code is quite easily understood.

My suggestion, if this approach/format looks okay, would be to add this to all the classes in hf_api. There is also the option of tweaking the width parameter used by pformat.

An example of the output given by the old and new __repr__ are shown below.

old

ModelInfo: {
	modelId: flyswot/convnext-tiny-224_flyswot
	sha: c6d4b2138e10efeafef8f5305ce16270ca583618
	lastModified: 2022-04-05T16:08:35.000Z
	tags: ['pytorch', 'convnext', 'image-classification', 'dataset:image_folder', 'transformers', 'generated_from_trainer', 'model-index']
	pipeline_tag: image-classification
	siblings: [RepoFile(rfilename='.gitattributes'), RepoFile(rfilename='.gitignore'), RepoFile(rfilename='README.md'), RepoFile(rfilename='config.json'), RepoFile(rfilename='preprocessor_config.json'), RepoFile(rfilename='pytorch_model.bin'), RepoFile(rfilename='training_args.bin')]
	private: False
	author: flyswot
	config: {'architectures': ['ConvNextForImageClassification'], 'model_type': 'convnext'a
	id: flyswot/convnext-tiny-224_flyswot
	downloads: 984
	library_name: transformers
	model-index: [{'name': 'convnext-tiny-224_flyswot', 'results': [{'task': {'name': 'Image Classification', 'type': 'image-classification'}, 'dataset': {'name': 'image_folder', 'type': 'image_folder', 'args': 'default'}, 'metrics': [{'name': 'F1', 'type': 'f1', 'value': 0.9756290792360154}]}]}]
	cardData: {'tags': ['generated_from_trainer'], 'datasets': ['image_folder'], 'metrics': ['f1'], 'model-index': [{'name': 'convnext-tiny-224_flyswot', 'results': [{'task': {'name': 'Image Classification', 'type': 'image-classification'}, 'dataset': {'name': 'image_folder', 'type': 'image_folder', 'args': 'default'}, 'metrics': [{'name': 'F1', 'type': 'f1', 'value': 0.9756290792360154}]}]}]}
	transformersInfo: {'auto_model': 'AutoModelForImageClassification', 'pipeline_tag': 'image-classification', 'processor': 'AutoFeatureExtractor'}
}

new

ModelInfo: { 
        {'modelId': 'flyswot/convnext-tiny-224_flyswot',
         'sha': 'c6d4b2138e10efeafef8f5305ce16270ca583618',
         'lastModified': '2022-04-05T16:08:35.000Z',
         'tags': ['pytorch',
                  'convnext',
                  'image-classification',
                  'dataset:image_folder',
                  'transformers',
                  'generated_from_trainer',
                  'model-index'],
         'pipeline_tag': 'image-classification',
         'siblings': [RepoFile(rfilename='.gitattributes'),
                      RepoFile(rfilename='.gitignore'),
                      RepoFile(rfilename='README.md'),
                      RepoFile(rfilename='config.json'),
                      RepoFile(rfilename='preprocessor_config.json'),
                      RepoFile(rfilename='pytorch_model.bin'),
                      RepoFile(rfilename='training_args.bin')],
         'private': False,
         'author': 'flyswot',
         'config': {'architectures': ['ConvNextForImageClassification'],
                    'model_type': 'convnext'},
         'id': 'flyswot/convnext-tiny-224_flyswot',
         'downloads': 984,
         'library_name': 'transformers',
         'model-index': [{'name': 'convnext-tiny-224_flyswot',
                          'results': [{'task': {'name': 'Image Classification',
                                                'type': 'image-classification'},
                                       'dataset': {'name': 'image_folder',
                                                   'type': 'image_folder',
                                                   'args': 'default'},
                                       'metrics': [{'name': 'F1',
                                                    'type': 'f1',
                                                    'value': 0.9756290792360154}]}]}],
         'cardData': {'tags': ['generated_from_trainer'],
                      'datasets': ['image_folder'],
                      'metrics': ['f1'],
                      'model-index': [{'name': 'convnext-tiny-224_flyswot',
                                       'results': [{'task': {'name': 'Image Classification',
                                                             'type': 'image-classification'},
                                                    'dataset': {'name': 'image_folder',
                                                                'type': 'image_folder',
                                                                'args': 'default'},
                                                    'metrics': [{'name': 'F1',
                                                                 'type': 'f1',
                                                                 'value': 0.9756290792360154}]}]}]},
         'transformersInfo': {'auto_model': 'AutoModelForImageClassification',
                              'pipeline_tag': 'image-classification',
                              'processor': 'AutoFeatureExtractor'}}
	}

[davanstrien]

cc @nateraw @adrinjalali @LysandreJik

[Wauplin]

Hi @davanstrien thanks for your PR. I want to follow-up on it. Is it still a Draft/WIP work or it's ready to review/merge ?
(cc @julien-c @LysandreJik as well)

[davanstrien]

Hi @davanstrien thanks for your PR. I want to follow-up on it. Is it still a Draft/WIP work or it's ready to review/merge ? (cc @julien-c @LysandreJik as well)

If everyone is happy with the suggested approach, then I would suggest adding this updated __repr__ to all the classes in hf_api.py. Once that is done this PR should be ready for final review/to be merged.

[Wauplin]

I quickly looked at hf_api.py and as you suggested it makes sense to adapt the __repr__ for all of them (RepoFile ModelInfo, DatasetInfo, SpaceInfo and MetricInfo). I think it would also be good to group everything in a single abstract class (implementing only __repr__) and give it by inheritance instead of the current duplicated code.

@davanstrien Since you already started working on it, would you mind taking care of it ? Thanks in advance :)

[davanstrien]

I quickly looked at hf_api.py and as you suggested it makes sense to adapt the __repr__ for all of them (RepoFile ModelInfo, DatasetInfo, SpaceInfo and MetricInfo). I think it would also be good to group everything in a single abstract class (implementing only __repr__) and give it by inheritance instead of the current duplicated code.

@davanstrien Since you already started working on it, would you mind taking care of it ? Thanks in advance :)

Sure, I'm super busy at the moment so I likely won't be able to look at this until mid-September.

[Wauplin]

Ok thanks, that's fine 👍

[HuggingFaceDocBuilderDev]

The documentation is not available anymore as the PR was closed or merged.

[davanstrien]

@Wauplin I've used a Mixin to implement this new __repr__ format. Let me know if you have something else in mind :)
I don't think the failing tests are because of these changes but let me know if I need to do anything to fix these.

[Wauplin]

Thanks for completing this @davanstrien ! Failing tests are indeed not related. I'm merging the PR :)