ADD: self-attention model #281

michaelteboul · 2025-11-03T13:28:57Z

Description of the goal of the PR

This pull request introduces a significant new recommendation model, the SelfAttentionModel, which leverages attention mechanisms and user embeddings for more personalized recommendations. It also refactors the core data structures to properly handle user identities, enabling the development of user-aware models. The changes include updates to existing models to integrate user data, improvements in evaluation methodologies with new metrics, and enhancements to data generation and loading processes to support these new capabilities. The overall aim is to expand the framework's ability to build and evaluate advanced recommendation systems.

Checklist before requesting a review

I have commented my code, particularly in hard-to-understand areas
I have typed my code
I have created / updated the docstrings
I have updated the README, if relevant
I have updated the requirements files if a new package is used
I have tested my code
The CI pipeline passes
I have performed a self-review of my code

… movie data retrieval

gemini-code-assist · 2025-11-03T13:29:28Z

Summary of Changes

Hello @michaelteboul, I'm Gemini Code Assist¹! I'm currently reviewing this pull request and will post my feedback shortly. In the meantime, here's a summary to help you and other reviewers quickly get up to speed!

This pull request introduces a significant new recommendation model, the SelfAttentionModel, which leverages attention mechanisms and user embeddings for more personalized recommendations. It also refactors the core data structures to properly handle user identities, enabling the development of user-aware models. The changes include updates to existing models to integrate user data, improvements in evaluation methodologies with new metrics, and enhancements to data generation and loading processes to support these new capabilities. The overall aim is to expand the framework's ability to build and evaluate advanced recommendation systems.

Highlights

New Self-Attention Model Introduced: A new SelfAttentionModel has been added, designed for sequential item recommendation. This model incorporates both short-term preferences (using an attention mechanism over recent basket items) and long-term preferences (using user and item embeddings). It includes methods for computing short-term and long-term distances, masked attention, and a combined loss function with ridge regularization.
User ID Integration in TripDataset: The Trip and TripDataset classes have been extended to include user_id. This allows for tracking individual user behavior, which is essential for user-specific recommendation models. The data generation methods in SyntheticDataGenerator were also updated to support user profiles and generate user-specific baskets.
Enhanced Evaluation Metrics: New evaluation metrics, hit_rate (Hit@K) and mrr (Mean Reciprocal Rank), have been implemented in the AleaCarta and SelfAttentionModel classes. These metrics are crucial for assessing the performance of recommendation systems, with logic to exclude already purchased items from recommendations.
Model Refinements and Timing: Existing models like AleaCarta and Shopper have been updated to support the new user_batch parameter. Performance timing (time.perf_counter()) has been integrated into AleaCarta and BaseBasketModel's training steps to monitor execution time. The loss normalization in AleaCarta's compute_batch_loss was adjusted, and Shopper now explicitly handles 'end-of-basket' items with dedicated parameters.
Dataset Handling Improvements: The bakery dataset loading utility has been streamlined, removing an outdated version parameter and simplifying file extraction. The synthetic_dataset module was moved and updated to generate user-specific data, aligning with the new user ID features.

Using Gemini Code Assist

The full guide for Gemini Code Assist can be found on our documentation page, here are some quick tips.

Invoking Gemini

You can request assistance from Gemini at any point by creating a comment using either /gemini <command> or @gemini-code-assist <command>. Below is a summary of the supported commands on the current page.

Feature	Command	Description
Code Review	`/gemini review`	Performs a code review for the current pull request in its current state.
Pull Request Summary	`/gemini summary`	Provides a summary of the current pull request in its current state.
Comment	@gemini-code-assist	Responds in comments when explicitly tagged, both in pull request comments and review comments.
Help	`/gemini help`	Displays a list of available commands.

Customization

To customize Gemini Code Assist for GitHub experience, repository maintainers can create a configuration file and/or provide a custom code review style guide (such as PEP-8 for Python) by creating and adding files to a .gemini/ folder in the base of the repository. Detailed instructions can be found here.

Limitations & Feedback

Gemini Code Assist may make mistakes. Please leave feedback on any instances where its feedback is incorrect or counter productive. You can react with 👍 and 👎 on @gemini-code-assist comments. If you're interested in giving your feedback about your experience with Gemini Code Assist for Github and other Google products, sign up here.

You can also get AI-powered code generation, chat, as well as code reviews directly in the IDE at no cost with the Gemini Code Assist IDE Extension.

Review the Privacy Notices, Generative AI Prohibited Use Policy, Terms of Service, and learn how to configure Gemini Code Assist in GitHub here. Gemini can make mistakes, so double check it and use code with caution. ↩

choice_learn/basket_models/datasets/bakery.py

for more information, see https://pre-commit.ci

gemini-code-assist

Code Review

This pull request introduces a new SelfAttentionModel, adds user-specific features, and refactors several parts of the codebase. The changes are substantial and add valuable new capabilities. However, there are several areas that require attention before merging. The code includes temporary profiling and debugging artifacts (e.g., print statements, time.perf_counter() calls) that should be removed. I've also identified some potential design issues, such as a base class containing logic specific to its children, and inconsistencies in data representation (e.g., the checkout item ID). Additionally, there are potential bugs related to missing security checks during file extraction, lack of bounds checking in data processing, and incorrect validation procedures. My review provides specific suggestions to address these points and improve the overall quality and robustness of the code.

github-actions · 2025-11-03T13:38:00Z

Coverage Report for Python 3.11

File	Stmts	Miss	Cover	Missing
choice_learn
__init__.py	2	0	100%
tf_ops.py	62	1	98%	283
choice_learn/basket_models
__init__.py	4	0	100%
alea_carta.py	148	22	85%	86–90, 92–96, 98–102, 106, 109, 131, 159, 308, 431–455
base_basket_model.py	235	27	89%	111–112, 123, 141, 185, 255, 377, 485, 585–587, 676, 762, 772, 822–836, 897–900, 940–941
basic_attention_model.py	89	4	96%	424, 427, 433, 440
self_attention_model.py	133	9	93%	71, 73, 75, 450–454, 651
shopper.py	184	9	95%	130, 159, 325, 345, 360, 363, 377, 489, 618
choice_learn/basket_models/data
__init__.py	2	0	100%
basket_dataset.py	190	30	84%	74–77, 295–297, 407, 540–576, 636, 658–661, 700–705, 790–801, 849
preprocessing.py	94	78	17%	43–45, 128–364
choice_learn/basket_models/datasets
__init__.py	3	0	100%
bakery.py	38	3	92%	47, 51, 61
synthetic_dataset.py	81	6	93%	62, 194–199, 247
choice_learn/basket_models/utils
__init__.py	0	0	100%
permutation.py	22	1	95%	37
choice_learn/data
__init__.py	3	0	100%
choice_dataset.py	649	33	95%	198, 250, 283, 421, 463–464, 589, 724, 738, 840, 842, 937, 957–961, 1140, 1159–1161, 1179–1181, 1209, 1214, 1223, 1240, 1281, 1293, 1307, 1346, 1361, 1366, 1395, 1408, 1443–1444
indexer.py	241	23	90%	20, 31, 45, 60–67, 202–204, 219–230, 265, 291, 582
storage.py	161	6	96%	22, 33, 51, 56, 61, 71
store.py	72	72	0%	3–275
choice_learn/datasets
__init__.py	4	0	100%
base.py	400	5	99%	42–43, 153–154, 714
expedia.py	102	83	19%	37–301
tafeng.py	49	0	100%
choice_learn/datasets/data
__init__.py	0	0	100%
choice_learn/models
__init__.py	14	2	86%	15–16
base_model.py	303	21	93%	144, 186, 283, 291, 297, 306, 346, 353, 382, 401, 432–433, 442–443, 544, 546, 562, 566, 568, 691–692
baseline_models.py	49	0	100%
conditional_logit.py	269	26	90%	49, 52, 54, 85, 88, 91–95, 98–102, 136, 206, 212–216, 351, 388, 445, 520–526, 651, 685, 822, 826
halo_mnl.py	124	2	98%	186, 374
latent_class_base_model.py	286	39	86%	55–61, 273–279, 288, 325–330, 497–500, 605, 624, 665–701, 715, 720, 751–752, 774–775, 869–870, 974
latent_class_mnl.py	62	6	90%	257–261, 296
learning_mnl.py	67	3	96%	157, 182, 188
nested_logit.py	291	12	96%	55, 77, 160, 269, 351, 484, 530, 600, 679, 848, 900, 904
reslogit.py	132	6	95%	285, 360, 369, 374, 382, 432
rumnet.py	236	3	99%	748–751, 982
simple_mnl.py	139	6	96%	167, 275, 347, 355, 357, 359
tastenet.py	94	3	97%	142, 180, 188
choice_learn/toolbox
__init__.py	0	0	100%
assortment_optimizer.py	27	6	78%	28–30, 93–95, 160–162
gurobi_opt.py	238	238	0%	3–675
or_tools_opt.py	230	11	95%	103, 107, 296–305, 315, 319, 607, 611
choice_learn/utils
metrics.py	85	43	49%	74, 126–130, 147–166, 176, 190–199, 211–232, 242
TOTAL	5614	839	85%

Tests	Skipped	Failures	Errors	Time
221	0 💤	1 ❌	0 🔥	8m 39s ⏱️

github-actions · 2025-11-03T13:38:01Z

Coverage Report for Python 3.9

File	Stmts	Miss	Cover	Missing
choice_learn
__init__.py	2	0	100%
tf_ops.py	62	1	98%	283
choice_learn/basket_models
__init__.py	4	0	100%
alea_carta.py	148	22	85%	86–90, 92–96, 98–102, 106, 109, 131, 159, 308, 431–455
base_basket_model.py	235	27	89%	111–112, 123, 141, 185, 255, 377, 485, 585–587, 676, 762, 772, 822–836, 897–900, 940–941
basic_attention_model.py	89	4	96%	424, 427, 433, 440
self_attention_model.py	133	9	93%	71, 73, 75, 450–454, 651
shopper.py	184	9	95%	130, 159, 325, 345, 360, 363, 377, 489, 618
choice_learn/basket_models/data
__init__.py	2	0	100%
basket_dataset.py	190	30	84%	74–77, 295–297, 407, 540–576, 636, 658–661, 700–705, 790–801, 849
preprocessing.py	94	78	17%	43–45, 128–364
choice_learn/basket_models/datasets
__init__.py	3	0	100%
bakery.py	38	3	92%	47, 51, 61
synthetic_dataset.py	81	6	93%	62, 194–199, 247
choice_learn/basket_models/utils
__init__.py	0	0	100%
permutation.py	22	1	95%	37
choice_learn/data
__init__.py	3	0	100%
choice_dataset.py	649	33	95%	198, 250, 283, 421, 463–464, 589, 724, 738, 840, 842, 937, 957–961, 1140, 1159–1161, 1179–1181, 1209, 1214, 1223, 1240, 1281, 1293, 1307, 1346, 1361, 1366, 1395, 1408, 1443–1444
indexer.py	241	23	90%	20, 31, 45, 60–67, 202–204, 219–230, 265, 291, 582
storage.py	161	6	96%	22, 33, 51, 56, 61, 71
store.py	72	72	0%	3–275
choice_learn/datasets
__init__.py	4	0	100%
base.py	400	5	99%	42–43, 153–154, 714
expedia.py	102	83	19%	37–301
tafeng.py	49	0	100%
choice_learn/datasets/data
__init__.py	0	0	100%
choice_learn/models
__init__.py	14	2	86%	15–16
base_model.py	303	21	93%	144, 186, 283, 291, 297, 306, 346, 353, 382, 401, 432–433, 442–443, 544, 546, 562, 566, 568, 691–692
baseline_models.py	49	0	100%
conditional_logit.py	269	26	90%	49, 52, 54, 85, 88, 91–95, 98–102, 136, 206, 212–216, 351, 388, 445, 520–526, 651, 685, 822, 826
halo_mnl.py	124	2	98%	186, 374
latent_class_base_model.py	286	39	86%	55–61, 273–279, 288, 325–330, 497–500, 605, 624, 665–701, 715, 720, 751–752, 774–775, 869–870, 974
latent_class_mnl.py	62	6	90%	257–261, 296
learning_mnl.py	67	3	96%	157, 182, 188
nested_logit.py	291	12	96%	55, 77, 160, 269, 351, 484, 530, 600, 679, 848, 900, 904
reslogit.py	132	6	95%	285, 360, 369, 374, 382, 432
rumnet.py	236	3	99%	748–751, 982
simple_mnl.py	139	6	96%	167, 275, 347, 355, 357, 359
tastenet.py	94	3	97%	142, 180, 188
choice_learn/toolbox
__init__.py	0	0	100%
assortment_optimizer.py	27	6	78%	28–30, 93–95, 160–162
gurobi_opt.py	236	236	0%	3–675
or_tools_opt.py	230	11	95%	103, 107, 296–305, 315, 319, 607, 611
choice_learn/utils
metrics.py	85	43	49%	74, 126–130, 147–166, 176, 190–199, 211–232, 242
TOTAL	5612	837	85%

Tests	Skipped	Failures	Errors	Time
221	0 💤	0 ❌	0 🔥	7m 51s ⏱️

github-actions · 2025-11-03T13:38:09Z

Coverage Report for Python 3.10

File	Stmts	Miss	Cover	Missing
choice_learn
__init__.py	2	0	100%
tf_ops.py	62	1	98%	283
choice_learn/basket_models
__init__.py	4	0	100%
alea_carta.py	148	22	85%	86–90, 92–96, 98–102, 106, 109, 131, 159, 308, 431–455
base_basket_model.py	235	27	89%	111–112, 123, 141, 185, 255, 377, 485, 585–587, 676, 762, 772, 822–836, 897–900, 940–941
basic_attention_model.py	89	4	96%	424, 427, 433, 440
self_attention_model.py	133	9	93%	71, 73, 75, 450–454, 651
shopper.py	184	9	95%	130, 159, 325, 345, 360, 363, 377, 489, 618
choice_learn/basket_models/data
__init__.py	2	0	100%
basket_dataset.py	190	30	84%	74–77, 295–297, 407, 540–576, 636, 658–661, 700–705, 790–801, 849
preprocessing.py	94	78	17%	43–45, 128–364
choice_learn/basket_models/datasets
__init__.py	3	0	100%
bakery.py	38	3	92%	47, 51, 61
synthetic_dataset.py	81	6	93%	62, 194–199, 247
choice_learn/basket_models/utils
__init__.py	0	0	100%
permutation.py	22	1	95%	37
choice_learn/data
__init__.py	3	0	100%
choice_dataset.py	649	33	95%	198, 250, 283, 421, 463–464, 589, 724, 738, 840, 842, 937, 957–961, 1140, 1159–1161, 1179–1181, 1209, 1214, 1223, 1240, 1281, 1293, 1307, 1346, 1361, 1366, 1395, 1408, 1443–1444
indexer.py	241	23	90%	20, 31, 45, 60–67, 202–204, 219–230, 265, 291, 582
storage.py	161	6	96%	22, 33, 51, 56, 61, 71
store.py	72	72	0%	3–275
choice_learn/datasets
__init__.py	4	0	100%
base.py	400	5	99%	42–43, 153–154, 714
expedia.py	102	83	19%	37–301
tafeng.py	49	0	100%
choice_learn/datasets/data
__init__.py	0	0	100%
choice_learn/models
__init__.py	14	2	86%	15–16
base_model.py	303	21	93%	144, 186, 283, 291, 297, 306, 346, 353, 382, 401, 432–433, 442–443, 544, 546, 562, 566, 568, 691–692
baseline_models.py	49	0	100%
conditional_logit.py	269	26	90%	49, 52, 54, 85, 88, 91–95, 98–102, 136, 206, 212–216, 351, 388, 445, 520–526, 651, 685, 822, 826
halo_mnl.py	124	2	98%	186, 374
latent_class_base_model.py	286	39	86%	55–61, 273–279, 288, 325–330, 497–500, 605, 624, 665–701, 715, 720, 751–752, 774–775, 869–870, 974
latent_class_mnl.py	62	6	90%	257–261, 296
learning_mnl.py	67	3	96%	157, 182, 188
nested_logit.py	291	12	96%	55, 77, 160, 269, 351, 484, 530, 600, 679, 848, 900, 904
reslogit.py	132	6	95%	285, 360, 369, 374, 382, 432
rumnet.py	236	3	99%	748–751, 982
simple_mnl.py	139	6	96%	167, 275, 347, 355, 357, 359
tastenet.py	94	3	97%	142, 180, 188
choice_learn/toolbox
__init__.py	0	0	100%
assortment_optimizer.py	27	6	78%	28–30, 93–95, 160–162
gurobi_opt.py	238	238	0%	3–675
or_tools_opt.py	230	11	95%	103, 107, 296–305, 315, 319, 607, 611
choice_learn/utils
metrics.py	85	43	49%	74, 126–130, 147–166, 176, 190–199, 211–232, 242
TOTAL	5614	839	85%

Tests	Skipped	Failures	Errors	Time
221	0 💤	0 ❌	0 🔥	8m 32s ⏱️

github-actions · 2025-11-03T13:39:21Z

Coverage Report for Python 3.12

File	Stmts	Miss	Cover	Missing
choice_learn
__init__.py	2	0	100%
tf_ops.py	62	1	98%	283
choice_learn/basket_models
__init__.py	4	0	100%
alea_carta.py	148	22	85%	86–90, 92–96, 98–102, 106, 109, 131, 159, 308, 431–455
base_basket_model.py	235	27	89%	111–112, 123, 141, 185, 255, 377, 485, 585–587, 676, 762, 772, 822–836, 897–900, 940–941
basic_attention_model.py	89	4	96%	424, 427, 433, 440
self_attention_model.py	133	9	93%	71, 73, 75, 450–454, 651
shopper.py	184	9	95%	130, 159, 325, 345, 360, 363, 377, 489, 618
choice_learn/basket_models/data
__init__.py	2	0	100%
basket_dataset.py	190	30	84%	74–77, 295–297, 407, 540–576, 636, 658–661, 700–705, 790–801, 849
preprocessing.py	94	78	17%	43–45, 128–364
choice_learn/basket_models/datasets
__init__.py	3	0	100%
bakery.py	38	3	92%	47, 53, 61
synthetic_dataset.py	81	6	93%	62, 194–199, 247
choice_learn/basket_models/utils
__init__.py	0	0	100%
permutation.py	22	1	95%	37
choice_learn/data
__init__.py	3	0	100%
choice_dataset.py	649	33	95%	198, 250, 283, 421, 463–464, 589, 724, 738, 840, 842, 937, 957–961, 1140, 1159–1161, 1179–1181, 1209, 1214, 1223, 1240, 1281, 1293, 1307, 1346, 1361, 1366, 1395, 1408, 1443–1444
indexer.py	241	23	90%	20, 31, 45, 60–67, 202–204, 219–230, 265, 291, 582
storage.py	161	6	96%	22, 33, 51, 56, 61, 71
store.py	72	72	0%	3–275
choice_learn/datasets
__init__.py	4	0	100%
base.py	400	5	99%	42–43, 153–154, 714
expedia.py	102	83	19%	37–301
tafeng.py	49	0	100%
choice_learn/datasets/data
__init__.py	0	0	100%
choice_learn/models
__init__.py	14	2	86%	15–16
base_model.py	303	21	93%	144, 186, 283, 291, 297, 306, 346, 353, 382, 401, 432–433, 442–443, 544, 546, 562, 566, 568, 691–692
baseline_models.py	49	0	100%
conditional_logit.py	269	26	90%	49, 52, 54, 85, 88, 91–95, 98–102, 136, 206, 212–216, 351, 388, 445, 520–526, 651, 685, 822, 826
halo_mnl.py	124	18	85%	186, 341, 360, 364–380
latent_class_base_model.py	286	39	86%	55–61, 273–279, 288, 325–330, 497–500, 605, 624, 665–701, 715, 720, 751–752, 774–775, 869–870, 974
latent_class_mnl.py	62	6	90%	257–261, 296
learning_mnl.py	67	3	96%	157, 182, 188
nested_logit.py	291	12	96%	55, 77, 160, 269, 351, 484, 530, 600, 679, 848, 900, 904
reslogit.py	132	6	95%	285, 360, 369, 374, 382, 432
rumnet.py	236	3	99%	748–751, 982
simple_mnl.py	139	6	96%	167, 275, 347, 355, 357, 359
tastenet.py	94	3	97%	142, 180, 188
choice_learn/toolbox
__init__.py	0	0	100%
assortment_optimizer.py	27	6	78%	28–30, 93–95, 160–162
gurobi_opt.py	238	238	0%	3–675
or_tools_opt.py	230	11	95%	103, 107, 296–305, 315, 319, 607, 611
choice_learn/utils
metrics.py	85	43	49%	74, 126–130, 147–166, 176, 190–199, 211–232, 242
TOTAL	5614	855	85%

Tests	Skipped	Failures	Errors	Time
221	0 💤	1 ❌	0 🔥	9m 23s ⏱️

Merge sub_branch for attention model with choice-learn

for more information, see https://pre-commit.ci

choice_learn/basket_models/datasets/bakery.py

choice_learn/basket_models/data/basket_dataset.py

choice_learn/basket_models/self_attention_model.py

for more information, see https://pre-commit.ci

choice_learn/basket_models/self_attention_model.py

choice_learn/basket_models/data/basket_dataset.py

choice_learn/basket_models/alea_carta.py

choice_learn/basket_models/base_basket_model.py

choice_learn/basket_models/basic_attention_model.py

choice_learn/utils/metrics.py

tests/integration_tests/basket_models/test_self-attention.py

VincentAuriau

quelques petits commentaires =)

michaelteboul added 6 commits October 10, 2025 11:22

Re install submodule changes

e91b95f

Add user and correct hit_rate

b1beae3

Add MRR calculation to AleaCarta; enhance TripDataset with sequential…

c58d8ae

… movie data retrieval

Update

a000d60

Add SelfAttentionModel implementation and corresponding tests

dcc77a5

Add self attention notebook

20b8eca

michaelteboul requested a review from VincentAuriau as a code owner November 3, 2025 13:28

github-advanced-security bot found potential problems Nov 3, 2025

View reviewed changes

choice_learn/basket_models/datasets/bakery.py Fixed Show fixed Hide fixed

[pre-commit.ci] auto fixes from pre-commit.com hooks

9089599

for more information, see https://pre-commit.ci

gemini-code-assist bot reviewed Nov 3, 2025

View reviewed changes

VincentAuriau changed the title ~~Sub branch~~ ADD: self-attention model Nov 5, 2025

michaelteboul and others added 4 commits November 5, 2025 14:19

Merge remote-tracking branch 'origin' into sub_branch

2bf90aa

Merge sub_branch for attention model with choice-learn

Switching folder

7dfa12a

Update

29fcf73

[pre-commit.ci] auto fixes from pre-commit.com hooks

87396e0

for more information, see https://pre-commit.ci

github-advanced-security bot found potential problems Nov 13, 2025

View reviewed changes

choice_learn/basket_models/datasets/bakery.py Fixed Show fixed Hide fixed

VincentAuriau reviewed Nov 13, 2025

View reviewed changes

michaelteboul and others added 7 commits November 13, 2025 16:26

Upgrade Aleacarta

1a9c564

Upgrade base_basket

00772bd

[pre-commit.ci] auto fixes from pre-commit.com hooks

8aef128

for more information, see https://pre-commit.ci

Upgrade self_attention

eb2380d

Upgrade synthetic_dataset

3bf4b0e

Upgrade basket_dataset

e8be87c

[pre-commit.ci] auto fixes from pre-commit.com hooks

44c5f56

for more information, see https://pre-commit.ci