Chore: Documentation and input / output cols #715
Conversation
| @@ -135,7 +154,7 @@ def _get_data_dict(self) -> Dict: | |||
| C.epoch.value: [self.epoch] * batch_size, | |||
| } | |||
| if self.split == Split.inference: | |||
| data["inference_name"] = [self.inference_name] * batch_size | |||
| data[C.inference_name.value] = [self.inference_name] * batch_size | |||
There was a problem hiding this comment.
Do we need a .value here? I think it works simply, too?
| data[C.inference_name.value] = [self.inference_name] * batch_size | |
| data[C.inference_name] = [self.inference_name] * batch_size |
There was a problem hiding this comment.
Yes we do! This creates a dict that later gets cast to a vaex DataFrame, which can't have enum as the keys to the dict
The following errors:
from enum import Enum
import vaex
class MyType(str, Enum):
a = "a"
b = "b"
c = "c"
df = vaex.from_dict(
{
MyType.a: [1, 2, 3],
MyType.b: [4, 5, 6],
MyType.c: [7, 8, 9],
}
)| token_deps = "token_deps" | ||
| token_gold_probs = "token_gold_probs" | ||
| # Mypy complained about split as an attribute, so we use `split_` | ||
| split_ = "split" | ||
| epoch = "epoch" | ||
| inference_name = "inference_name" |
There was a problem hiding this comment.
For line 157 in dataquality/loggers/model_logger/seq2seq.py
There was a problem hiding this comment.
My question was why call it inference name? Name as a column name is too generic and doesn't feel like that's what we should be getting?
Codecov Report
@@ Coverage Diff @@
## main #715 +/- ##
=======================================
Coverage 89.69% 89.70%
=======================================
Files 166 166
Lines 13320 13323 +3
=======================================
+ Hits 11948 11951 +3
Misses 1372 1372
... and 2 files with indirect coverage changes 📣 We’re building smart automated test selection to slash your CI/CD build times. Learn more |
Small cleanup leading up to S2S sprint during onsite week
Small cleanup leading up to S2S sprint during onsite week