Skip to content

SASRec

Self-Attentive Sequential Recommendation.

Overview

SASRec (Self-Attentive Sequential Recommendation) is a baseline model that uses self-attention mechanisms to capture user behavior patterns from sequential interaction data.

Architecture

User Sequence: [item_1, item_2, ..., item_n]
       ↓
Item Embeddings
       ↓
Positional Encoding
       ↓
Transformer Encoder (Self-Attention)
       ↓
Next Item Prediction

Key Components

  • Item Embedding: Learnable embeddings for all items
  • Positional Encoding: Learnable position embeddings
  • Self-Attention Layers: Multi-head attention with causal masking
  • Prediction Head: Dot product with item embeddings

Configuration

# config/sasrec/amazon.gin

train.epochs = 200
train.batch_size = 128
train.learning_rate = 1e-3
train.max_seq_len = 50

# Model architecture
train.hidden_dim = 64
train.num_heads = 2
train.num_layers = 2
train.dropout = 0.2

Training

# Train on Amazon Beauty
python genrec/trainers/sasrec_trainer.py config/sasrec/amazon.gin

# Train on other datasets
python genrec/trainers/sasrec_trainer.py config/sasrec/amazon.gin --split sports
python genrec/trainers/sasrec_trainer.py config/sasrec/amazon.gin --split toys

Evaluation Metrics

  • Recall@K: Proportion of relevant items in top-K recommendations
  • NDCG@K: Normalized Discounted Cumulative Gain at K

Benchmark Results

Amazon 2014 Beauty

Model R@5 R@10 N@5 N@10
SASRec 0.0469 0.0688 0.0305 0.0375

Model API

from genrec.models import SASRec

model = SASRec(
    num_items=10000,
    hidden_dim=64,
    num_heads=2,
    num_layers=2,
    max_seq_len=50,
    dropout=0.2,
)

# Forward pass
logits = model(item_ids, attention_mask)

Reference