exasol / transformers-extension

An Exasol extension for using state-of-the-art pretrained machine learning models via the Hugging Face Transformers API.
MIT License
2 stars 2 forks source link

Code cleanup and comments for better readability #135

Open MarleneKress79789 opened 11 months ago

MarleneKress79789 commented 11 months ago

Some things are rather complex but not explained. Some comments in the code would benefit new developers adding or reviewing code. Also it might be helpful to move some snippets to a more logical place.

### Tasks
- [ ] base_model_udf:  file does switch of the model, and scan for checking if shared model parameters(shared by all udfs, as oposed to task specific parameters) changed
- [ ] base_model_udf: mention names of functions that need implementing(probs also in dev docu)
- [ ] constants.py: add comment about constants for template rendering, and/or change name of file
- [ ] deployment_utils: move global params to slc deployer? and then pass to here deployment_utils
- [ ] dataframe_operations: sort_cell_values: for cells/column entries that contain comma separated lists
- [ ] TemporaryDirectoryFactory: add comment about class so can be mocked for tests
- [ ] scripts_deployer: add commnt that this deploys all existing udfs of transformers extension
- [ ] [load_models](https://github.com/exasol/transformers-extension/blob/1321af15f1cf0e9ac68ca12c02a3d6e4318a2cb1/exasol_transformers_extension/udfs/models/base_model_udf.py#L187) function description mentions a ticket that has since been closed
- [ ] mention in docu  in theory the alias could be changed, but also need to change it  [here](https://github.com/exasol/transformers-extension/blob/main/language_container/exasol_transformers_extension_container/flavor_base/language_definition)?