sbrl-github/research-rainfallradar
1
0
Fork 0
mirror of https://github.com/sbrl/research-rainfallradar synced 2024-11-25 10:32:59 +00:00
Code Issues Projects Releases Packages Wiki Activity

train_predict/jsonl: don't argmax

Browse source 
I'm interested inthe raw values
This commit is contained in:
Starbeamrainbowlabs 2022-10-26 17:21:19 +01:00
parent 74f2cdb900
commit 33391eaf16
Signed by: sbrl
GPG key ID: 1BE5172E637709C2
1 changed files with 1 additions and 1 deletions
Show stats Download patch file Download diff file Expand all files Collapse all files

2
aimodel/src/subcommands/train_predict.py
View file

@ -152,7 +152,7 @@ def do_jsonl(args, ai, dataset, model_params):
i_batch = 0 i_batch = 0
for water_predict in water_predict_batch: for water_predict in water_predict_batch:
# [ width, height, softmax_probabilities ] → [ batch, width, height ] # [ width, height, softmax_probabilities ] → [ batch, width, height ]
water_predict = tf.math.argmax(water_predict, axis=-1) # water_predict = tf.math.argmax(water_predict, axis=-1)
# [ width, height ] # [ width, height ]
water_actual = tf.squeeze(water_actual_batch[i_batch]) water_actual = tf.squeeze(water_actual_batch[i_batch])