SALMUBench: A Benchmark for Sensitive Association-Level Multimodal Unlearning

SALMUBench is the official evaluation dataset for a CVPR 2026 benchmark on multimodal unlearning. The dataset, authored by cvc-mmu, is designed to assess methods for removing sensitive associations from models. It was last updated on March 30, 2026.

Use Cases

• Benchmarking the performance of unlearning algorithms based on the described forget and utility metrics.
• Evaluating a model's ability to forget sensitive associations based on the benchmark's defined workflow.
• Comparing a compromised model against a clean reference model as outlined in the dataset's typical use case.
• Researching the trade-offs between forgetting efficacy and model utility in multimodal systems.

Strengths

• Dataset is the official benchmark for a CVPR 2026 paper, indicating academic relevance.
• Last update timestamp is precisely recorded as 2026-03-30 07:28:11.

Limitations

• Column-level documentation is absent; field semantics must be inferred after download.
• Row count is unknown, which may limit suitability assessment.
• Description metadata is limited; actual data quality requires manual inspection after download.

Provenance

Source

cvc-mmu on Hugging Face.

Collection Method

Created as an evaluation benchmark for the SALMUBench paper.

Time Range

Associated with research for CVPR 2026.

Freshness

Last updated 2026-03-30 07:28:11; freshness should be verified.

Geography

null