📊 AI: AI epoch adding regressions code

[veronikasamborska1994]

hey @Marigold this is the remainder of the code that I accidentally pushed to master that caused the bug! it's a monthly update so most of these have been reviewed several times before except this new bit of code with regressions

[owidbot]

Quick links (staging server):

Site	Admin	Wizard

Login: ssh owid@staging-site-ai-epoch-sep

chart-diff: ✅

• 20/20 reviewed charts
• Modified: 20/20
• New: 0/0

data-diff: ✅ No differences found

+ Dataset garden/artificial_intelligence/2024-10-01/epoch_aggregates_affiliation
+ + Table epoch_aggregates_affiliation
+   + Column yearly_count
+   + Column cumulative_count
+ Dataset garden/artificial_intelligence/2024-10-01/epoch_aggregates_domain
+ + Table epoch_aggregates_domain
+   + Column yearly_count
+   + Column cumulative_count
+ Dataset garden/artificial_intelligence/2024-10-01/epoch_compute_intensive
+ + Table epoch_compute_intensive
+   + Column publication_date
+   + Column domain
+   + Column parameters
+   + Column training_dataset_size__datapoints
+   + Column training_computation_petaflop
+ Dataset garden/artificial_intelligence/2024-10-01/epoch_compute_intensive_countries
+ + Table epoch_compute_intensive_countries
+   + Column yearly_count
+   + Column cumulative_count
+ Dataset garden/artificial_intelligence/2024-10-01/epoch_compute_intensive_domain
+ + Table epoch_compute_intensive_domain
+   + Column yearly_count
+   + Column cumulative_count
+ Dataset garden/artificial_intelligence/2024-10-01/epoch_regressions
+ + Table epoch_regressions
+   + Column training_computation_petaflop
+   + Column parameters
+   + Column training_dataset_size__datapoints
+   + Column publication_date
+   + Column domain
+   + Column organization_categorization


Legend: +New  ~Modified  -Removed  =Identical  Details
Hint: Run this locally with etl diff REMOTE data/ --include yourdataset --verbose --snippet

Automatically updated datasets matching weekly_wildfires|excess_mortality|covid|fluid|flunet|country_profile|garden/ihme_gbd/2019/gbd_risk are not included

Edited: 2024-10-04 19:55:43 UTC
Execution time: 3.89 seconds

[Marigold]

Code looks good! (What a paradox that deep learning systems are being analyzed using simple regression...)

Some charts look notably different and show "Circles sized by Population", is that expected?

[veronikasamborska1994]

thanks @Marigold! nicely spotted - I guess I was changing some things and population sizing is the default which I didn't remove!