Closed cgarciae closed 2 years ago
Since recently we've been using the following convention for the arguments for Metric's update function:
update
y_pred
preds
y_true
target
However, MeanSquareError and MeanAbsoluteError use the old convention:
MeanSquareError
MeanAbsoluteError
https://github.com/cgarciae/treex/blob/3aa54154e8a28dc25aba7dd79bac3b0c7e56e1ba/treex/metrics/mean_square_error.py#L64-L69 https://github.com/cgarciae/treex/blob/3aa54154e8a28dc25aba7dd79bac3b0c7e56e1ba/treex/metrics/mean_absolute_error.py#L64-L69
This is not a hard issue for Treex itself but Elegy does rely on this convention and it makes these metrics incompatible (poets-ai/elegy#216).
cc: @anvelezec @jmarrietar
Since recently we've been using the following convention for the arguments for Metric's
update
function:y_pred
=>preds
y_true
=>target
However,
MeanSquareError
andMeanAbsoluteError
use the old convention:https://github.com/cgarciae/treex/blob/3aa54154e8a28dc25aba7dd79bac3b0c7e56e1ba/treex/metrics/mean_square_error.py#L64-L69 https://github.com/cgarciae/treex/blob/3aa54154e8a28dc25aba7dd79bac3b0c7e56e1ba/treex/metrics/mean_absolute_error.py#L64-L69
This is not a hard issue for Treex itself but Elegy does rely on this convention and it makes these metrics incompatible (poets-ai/elegy#216).
cc: @anvelezec @jmarrietar