🔮 v0.2.0 - TODOs
- [ ] Add
Typer
CLI e.g.opentrain train ...
- [ ] Add
Dataset
validation before actually uploading aDataset
/File
to OpenAI. - [ ] Add
Dataset.from_datasets
,Dataset.to_datasets
, andDataset.to_records
. - [ ] Add
fsspec
support forDataset.from_file
, andDataset.to_file
. - [ ] Allow different input paths such as
pathlib.Path
oros.path
inDataset.from_file
. - [ ] Explore https://github.com/openai/openai-python/blob/c556584eff3b36c92278e6af62cfe02ebb68fb65/openai/api_resources/file.py#L218 to avoid uploading duplicated files to OpenAI.
- [ ] Add
Trainer.for_text_classification
,Trainer.for_question_answering
,Trainer.for_text_summarization
, and more if applicable. - [ ] Add
wandb
as an optional dependency for tracking fine-tune runs. - [ ] Explore automatically uploaded files to OpenAI after fine-tuning with
purpose='fine-tune-results'
. - [ ] Differentiate between both file-purposes
fine-tune
andfine-tune-results
.
Last update:
2023-05-01
Created: 2023-05-01
Created: 2023-05-01