Model Library Overview

Torch-RecHub provides a rich set of recommendation models covering ranking, matching, multi-task learning, and generative recommendation. All models are implemented with PyTorch and are easy to use and extend.

Library Structure

Models are organized by stage/task:

1. Ranking: fine-ranking models to predict CTR or preference scores.
2. Matching: retrieval models to fetch candidate items from large catalogs.
3. Multi-Task: models that jointly optimize related tasks.
4. Generative: models that generate personalized recommendations.

Model Selection Guide

Ranking Models

Model	Scenario	Highlights
WideDeep	Basic ranking	Combines linear (memorization) and deep (generalization)
DeepFM	Feature interactions matter	Captures low- and high-order feature crosses
DCN/DCNv2	Explicit feature crossing	Efficient high-order crosses
DIN	Dynamic user interest	Attention-based interest modeling
DIEN	Long sequence interest	Models evolving interests
BST	Sequence-focused	Transformer for sequential features
AutoInt	Auto feature interaction	Learns interaction patterns automatically

Matching Models

Model	Scenario	Highlights
DSSM	Text / semantic match	Two-tower mapping to a shared vector space
YoutubeDNN	Large-scale retrieval	Sequence-based deep retrieval
MIND	Multi-interest	Learns multiple user interests
GRU4Rec/SASRec	Sequential recommendation	Models recent behavior sequences
ComirecDR/ComirecSA	Controllable interests	Control number of interests

Multi-Task Models

Model	Scenario	Highlights
SharedBottom	Tasks highly related	Shared bottom network
MMOE	Task conflict exists	Multi-gate mixture-of-experts per task
PLE	Complex multi-task	Progressive layered extraction to reduce negative transfer
ESMM	Sample selection bias	Full-space modeling to mitigate bias
AITM	Task dependency	Adaptive information transfer between tasks

Generative Recommendation

Model	Scenario	Highlights
HSTU	Large-scale sequences	Hierarchical Sequential Transduction Units
HLLM	LLM-enhanced recommendation	Combines LLM semantic understanding

Documentation Navigation

• Ranking models: detailed principles, usage, and parameters.
  See ranking docs
• Matching models: detailed principles, usage, and parameters.
  See matching docs
• Multi-task models: detailed principles, usage, and parameters.
  See multi-task docs
• Generative models: principles, usage, and parameters.
  See generative docs

Usage Examples

# Ranking example
from torch_rechub.models.ranking import DeepFM
from torch_rechub.trainers import CTRTrainer

model = DeepFM(
    deep_features=deep_features,
    fm_features=fm_features,
    mlp_params={"dims": [256, 128], "dropout": 0.2}
)
trainer = CTRTrainer(model, optimizer_params={"lr": 0.001}, device="cuda:0")
trainer.fit(train_dataloader, val_dataloader)

# Matching example
from torch_rechub.models.matching import DSSM
from torch_rechub.trainers import MatchTrainer

model = DSSM(
    user_features=user_features,
    item_features=item_features,
    temperature=0.02,
    user_params={"dims": [256, 128, 64]},
    item_params={"dims": [256, 128, 64]}
)
trainer = MatchTrainer(model, mode=0, optimizer_params={"lr": 0.001}, device="cuda:0")
trainer.fit(train_dataloader)