Closed celsofranssa closed 3 years ago
Hi @Ceceu ,
It will require a little coding, but yes.
In general, code2seq computes a new "code vector" for every target subtoken. So if the model predicts "save bitmap to file", a different "code vector" is computed for each of the outputs [save, bitmap, to, file]
.
If you need a single vector, you can use this contexts_average
vector:
https://github.com/tech-srl/code2seq/blob/master/model.py#L414
Best, Uri
Closing due to inactivity, but feel free to re-open if you have additional questions.
Is possible to use code2seq to output the code vector?