Code Retrieval
#2
by
kevinlu1248
- opened
Is this model optimized for code retrieval? What about text-to-code retrieval?
This model was pre-trained with the standard BERT objectives (MLM+NSP), so it needs to be fine-tuned before being used for retrieval.
However, in preliminary experiments, we've found it to work kind of ok in theses tasks even without fine-tuning. Maybe this can be useful if you want to try it yourself: https://github.com/bigcode-project/bigcode-encoder/blob/master/embedding_sandbox.ipynb