Use of `top_k` in `hugie endpoint test` fails on some tasks?
Closed this issue · 0 comments
ivyleavedtoadflax commented
On a token classification task, this fails, on da text classification task it works. Probably best to remove.
glglt 2022-11-23T11:56:01.539Z 2022-11-23 11:56:01,539 | INFO | No custom pipeline found at /repository/handler.py
glglt 2022-11-23T11:56:01.539Z 2022-11-23 11:56:01,539 | INFO | Using device CPU
glglt 2022-11-23T11:56:01.539Z 2022-11-23 11:56:01,539 | INFO | Initializing model from directory:/repository
glglt 2022-11-23T11:56:04.781Z 2022-11-23 11:56:04,781 | INFO | Model initialized successfully
glglt 2022-11-23T11:56:52.281Z 2022-11-23 11:56:52,280 | ERROR | _sanitize_parameters() got an unexpected keyword argument 'top_k'
glglt 2022-11-23T11:57:30.265Z 2022-11-23 11:57:30,265 | ERROR | _sanitize_parameters() got an unexpected keyword argument 'top_k'
glglt 2022-11-23T12:01:39.858Z 2022-11-23 12:01:39,858 | ERROR | _sanitize_parameters() got an unexpected keyword argument 'top_k'