v1.0.0-develop.8

1.0.0-develop.8 (2021-11-04)

Bug Fixes

• fixed field scaling issue with set index (7248b02)

v1.0.0-develop.7

1.0.0-develop.7 (2021-11-03)

Bug Fixes

• giving proper arguments to DSPN encoder and decoder (55a2c7b)

v1.0.0-develop.6

1.0.0-develop.6 (2021-11-03)

Bug Fixes

• added DSPN imports (e12adfb)

v1.0.0-develop.5

1.0.0-develop.5 (2021-11-03)

Bug Fixes

• added loss calculation in DSPNAE (8038f4b)

v1.0.0-develop.4

1.0.0-develop.4 (2021-11-03)

Bug Fixes

• dimensionality of input and mask for DSPN (b421e14)

v1.0.0-develop.3

1.0.0-develop.3 (2021-11-03)

Features

• first integration Deep Set Prediction Network (DSPN) (b548d96)

v1.0.0-develop.2

1.0.0-develop.2 (2021-10-27)

Performance Improvements

• just downloading 3 pages instead of 1500 from IATI.cloud (f8d3a17)

v1.0.0-develop.1

1.0.0-develop.1 (2021-10-27)

Bug Fixes

• added experiment name (check mlflow lingo) to runs configs (88e0037)
• airflow tasks will be started only manually (6b25f47)
• allow large numpy array objects to be stored in the mongodb via gridfs (3706d92)
• attempt to fix missing remote artifacts logging (ccf006e)
• bugs in text field encoding (677b81b)
• epochs measurement now are collected again (a4f5f82)
• evolution video script was not using the correct prefix variable to generate the video filename (004ac73)
• fixed a lot of bugs in the newly-refactored object-oriented measurement collecting frameweork (cb3789d)
• making the model load the training/testing sets from gridfs (4b0fa07)
• the hidden layers weren't considered ed at all by a misunderstanding (e4678ae)

Features

• added mlflow_log parameter to Measurement object to allow (d29cbaa)
• conf for runs of item autoencoder on budget set with a large bottleneck (681d4e7)
• config file as input argument and system to log to files (cfbf123)
• creation date of tsets stored (934fe2c)
• deactivate download of specific rels when not needed (e978338)
• deep sets model: encoder (d8b2dbb)
• deep sets model: new module (6a5e77b)
• hostname-based config system (03fb84d)
• split encode, arrayfy, to_npa and to_tsets tasks according to the relationship they handle (665125b)
• variable-length batchsizes for deep sets (099be04)
• yaml config to run the item autoencoder on the result dataset (f79c7fa)

Performance Improvements

• only downloading the necessary fields (6c305bc)