not all tensors properly used by train
erikr opened this issue · 1 comments
erikr commented
What
see log:
2020-06-24 15:57:52,684 - tensor_generators:232 - INFO -
!!!!>~~~~~~~~~~~~ train_worker completed true epoch 2 ~~~~~~~~~~~~<!!!!
Aggregated information string:
Generator looped & shuffled over 11317 paths. Epoch: 2
11322 tensors were presented.
0 paths were skipped because they previously failed.
No errors raised.
Categorical TensorMap: sts_death has 7992 total examples.
Label no_sts_death 7698 examples, 96.32% of total.
Label sts_death 294 examples, 3.68% of total.
Why
tbd
How
tbd
Acceptance Criteria
train uses all tensors
StevenSong commented
Pretty sure this is solved by #345