DIGER Processed Data: Artifacts for Generative Recommendation

Processed artifacts and embeddings used by the DIGER (Differentiable Semantic IDs for Generative Recommendation) model. The files are provided by author junchenfu to facilitate reproduction of the paper's results, as small preprocessing differences may affect outcomes. The dataset was last updated on June 8, 2026.

Use Cases

• Reproducing recommendation results based on the provided processed artifacts.
• Benchmarking semantic ID generation methods based on the described preprocessing pipeline.
• Developing differentiable recommendation models based on the associated paper's methodology.

Strengths

• Artifacts are directly linked to the published DIGER paper, enabling direct research verification.
• The dataset is provided explicitly to reduce reproduction variance from preprocessing differences.

Limitations

• Column-level documentation is absent; field semantics must be inferred after download.
• Row count and data scale are unknown, which may limit suitability assessment.

Provenance

Source

huggingface

Collection Method

Processed artifacts from the DIGER model pipeline.

Freshness

Last updated 2026-06-08 16:20:46.