v0.2.2

What's Changed

• adding download from huggingface when the finetune_output_dir is empty by @simona-rc in #19
• Support inner-sourced model loading & publishing & loading from hub by @SagiPolaczek in #20
• Add tutorial and README badges by @SagiPolaczek in #18
• Cleaning notebook MT outputs by @SagiPolaczek in #22
• Open pytest for all tests by @SagiPolaczek in #26
• Allow config mismatches by @SagiPolaczek in #24
• TCR beta chain - epitope binding inference by @floccinauc in #25
• Initialize MammalConfig from path + minor refactoring by @SagiPolaczek in #27
• The prompt generator in TDI adds unneaded white space - this should fix that by @matanninio in #23
• inference for MoleculeNet fine-tuned models BBBP, TOXICITY, FDA_APPR by @simona-rc in #21
• Refactor molnet_infer function by @matanninio in #28
• add an option to supply a model path to load_model in the TCR task by @floccinauc in #29
• ci fix by @mosheraboh in #31
• separate model path and tokenizer path by @floccinauc in #30
• Keep metrics information on the same device by @SagiPolaczek in #32
• 'scalars' was misspelled as 'scalers', by @matanninio in #36
• Set default values when loading prediction head by @SagiPolaczek in #38
• (partially) Fix pre-trained LoRA loading by @SagiPolaczek in #35
• Cell type from gene expression by @matanninio in #33

New Contributors

• @simona-rc made their first contribution in #19
• @floccinauc made their first contribution in #25

Full Changelog: 0.2.1...v0.2.2