Powerful unsupervised domain adaptation method for dense retrieval. Requires only unlabeled corpus and yields massive improvement: "GPL: Generative Pseudo Labeling for Unsupervised Domain Adaptation of Dense Retrieval" https://arxiv.org/abs/2112.07577
The current version could not identify OOM error in loadable_by_sbert_oom, since OOM is also a runtime error and this loadable checking views all runtime errors as not loadable
The current version could not identify OOM error in
loadable_by_sbert_oom, since OOM is also a runtime error and this loadable checking views all runtime errors as not loadablemodified: gpl/toolkit/sbert.py: Raise OOM error (runtime error)modified: setup.py: Added pytestnew file: tests/unit/conftest.py: SBERT fixturenew file: tests/unit/test_sbert.py: Test OOM error case