Skip to content

Releases: sisinflab/warprec

v1.4.6

Choose a tag to compare

@MarcoWV MarcoWV released this 03 Jun 09:08

v1.4.6 (2026-06-03)

Bug Fixes

  • Changed python version dependency to be >= 3.12 version (8a96a03)

  • CL4SRec model operation vectorization (9af00e8)

  • ESASRec model operation vectorization (65961b3)

  • LinRec model Q and K scaling factor during normalization (d79c7ec)

  • PopDCL model now applies normalization correctly during predict (8e98cc3)

Documentation

  • Removed unused hyperparameter from CORE configuration example (c4888be)

Detailed Changes: v1.4.5...v1.4.6

v1.4.5

Choose a tag to compare

@MarcoWV MarcoWV released this 22 May 13:48

v1.4.5 (2026-05-22)

Bug Fixes

  • Added address as general configuration parameter (bac5b93)

  • Evaluation dataloaders will accept and correctly cache kwargs (1fa6e13)

  • Model extended name depends also on original model name to ensure that models with same parameters generate the same name extension (c364ddb)

Documentation

  • Added ray address to documentation (b4494a2)

Detailed Changes: v1.4.4...v1.4.5

v1.4.4

Choose a tag to compare

@MarcoWV MarcoWV released this 20 May 13:58

v1.4.4 (2026-05-20)

Bug Fixes

  • Fixed bug with Timestamp Slicing splitting strategy which corrupted dataset dimensions (81aa8b6)

  • WarpRec will now log internal alignment filtering on evaluation sets (f1e85ca)


Detailed Changes: v1.4.3...v1.4.4

v1.4.3

Choose a tag to compare

@MarcoWV MarcoWV released this 13 May 14:27

v1.4.3 (2026-05-13)

Bug Fixes

  • Model names will be created using coolname based on their parameter hash (b424f7a)

  • Trial name during training will be derived from the new naming convention (86f14da)

Chores

  • Added coolname dependency (d606015)

Detailed Changes: v1.4.2...v1.4.3

v1.4.2

Choose a tag to compare

@MarcoWV MarcoWV released this 06 May 14:04

v1.4.2 (2026-05-06)

Bug Fixes

  • Added concurrent trial estimation and placement group compute during trial allocation (3dabb4b)

  • Custom Ray Train trial ID to prevent race condition during trial allocation (8627f52)

  • Removed placement group strategy and delegating resource management to Ray Train (66c3520)

Chores

  • Added max_concurrent_trial configuration (7d1e548)

Detailed Changes: v1.4.1...v1.4.2

v1.4.1

Choose a tag to compare

@MarcoWV MarcoWV released this 05 May 13:17

v1.4.1 (2026-05-05)

Bug Fixes

  • Added warning message in the swarm pipeline (750e321)

  • Corrected PAAC model loss computation (074d11a)

Documentation

  • Added references to the new estimate pipeline (fcad7b1)

Detailed Changes: v1.4.0...v1.4.1

v1.4.0

Choose a tag to compare

@MarcoWV MarcoWV released this 04 May 08:48

v1.4.0 (2026-05-04)

Bug Fixes

  • Fixed PAAC model implementation to correctly reflect original paper formulation (057b727)

  • Removed max trial constraint (d0be3be)

  • RP3Beta model initialization (b1e002e)

Documentation

  • Added new models to the documentation (bbead3c)

Features

  • Added bsarec model implementation (af91531)

  • Added cl4srec model implementation (1e26bee)

  • Added duorec model implementation (21a7345)

  • Added esasrec model implementation (2390ce9)

  • Added iALS model implementation (29b583d)

  • Added iALS2008 model implementation (11b5eed)

  • New dataloader for same target sequential data (2da35bf)


Detailed Changes: v1.3.2...v1.4.0

v1.3.2

Choose a tag to compare

@MarcoWV MarcoWV released this 29 Apr 08:56

v1.3.2 (2026-04-29)

Bug Fixes

  • MACRMF parameter validation fixed (0f98503)

  • Temporal Decay KNN models parameter validation fixed (cbea62b)

  • Train and Swarm pipeline will assign a fallback value to number of GPUs for cuda models with no specified number of cuda devices per trial (5d2b4d7)

Documentation

  • Updated model taxonomy and counting (94a2c0f)

Detailed Changes: v1.3.1...v1.3.2

v1.3.1

Choose a tag to compare

@MarcoWV MarcoWV released this 23 Apr 08:06

v1.3.1 (2026-04-23)

Bug Fixes

  • Full evaluation will now iterate over only the test set, not the full set of users and items (505853a)

Detailed Changes: v1.3.0...v1.3.1

v1.3.0

Choose a tag to compare

@MarcoWV MarcoWV released this 22 Apr 09:05

v1.3.0 (2026-04-22)

Bug Fixes

  • Correct ASHA initialization (a9669ed)

  • Correctly initialize optimization metric for bayesian strategies (4b9299e)

  • Correctly loading custom modules on remote worker nodes (7e0b168)

  • Correctly set reproducibility after final evaluation model de-serialization (f855350)

  • Metrics will not be validated with config, allowing custom metrics to be correctly imported (d2867a0)

  • Serialization error of Pop model (64fb583)

  • Tuning process will now account for driver function resources during CPU-bound optimizations (f2c6938)

Documentation

  • Added ASHA scheduler example usage (b514b11)

  • Added context label in the configuration setting (134ef19)

  • Updated documentation with new models (290127e)

Features

  • Added ItemKNN-TD model implementation (6c01021)

  • Added MACRGCN model implementation (98bf4e8)

  • Added MACRMF model implementation (91d162a)

  • Added PAAC model implementation (c2f5c15)

  • Added PopDCL model implementation (0a182a0)

  • Added RecDCL model implementation (738404a)

  • Added SimGCL model implementation (abba348)

  • Added SimRec model implementation (37a7257)

  • Added STAN model implementation (79cc29a)

  • Added UserKNN-TD model implementation (3034ad5)

  • Codecarbon local logging will be saved in WarpRec experiment directory (6d8ffff)


Detailed Changes: v1.2.2...v1.3.0