SALMUBench: A Benchmark for Sensitive Association-Level Multimodal Unlearning

SALMUBench: A Benchmark for Sensitive Association-Level Multimodal Unlearning | Signal Canvas | ScienceToStartup

Page Freshness

Signal Canvas proof surface

Canonical route: /signal-canvas/salmubench-a-benchmark-for-sensitive-association-level-multimodal-unlearning

stale

Proof freshness: stale
Proof status: unverified
Display score: 7/10
Last proof check: 2026-03-30
Score updated: 2026-04-02
Score fresh until: 2026-05-02
References: 52
Source count: 8
Coverage: 50%

This page is showing the last landed evidence receipt and score bundle because the latest proof data is outside the freshness window.

Agent Handoff

Canonical ID salmubench-a-benchmark-for-sensitive-association-level-multimodal-unlearning | Route /signal-canvas/salmubench-a-benchmark-for-sensitive-association-level-multimodal-unlearning

REST example

curl https://sciencetostartup.com/api/v1/agent-handoff/signal-canvas/salmubench-a-benchmark-for-sensitive-association-level-multimodal-unlearning

MCP example

{
  "tool": "search_signal_canvas",
  "arguments": {
    "mode": "paper",
    "paper_ref": "salmubench-a-benchmark-for-sensitive-association-level-multimodal-unlearning",
    "query_text": "Summarize SALMUBench: A Benchmark for Sensitive Association-Level Multimodal Unlearning"
  }
}

source_context

{
  "surface": "signal_canvas",
  "mode": "paper",
  "query": "SALMUBench: A Benchmark for Sensitive Association-Level Multimodal Unlearning",
  "normalized_query": "2603.26316",
  "route": "/signal-canvas/salmubench-a-benchmark-for-sensitive-association-level-multimodal-unlearning",
  "paper_ref": "salmubench-a-benchmark-for-sensitive-association-level-multimodal-unlearning",
  "topic_slug": null,
  "benchmark_ref": null,
  "dataset_ref": null
}

Evidence Receipt

Route status: building

Claims: 12

References: 52

Proof: Verification pending

Freshness state: computing

Source paper: SALMUBench: A Benchmark for Sensitive Association-Level Multimodal Unlearning

PDF: https://arxiv.org/pdf/2603.26316v1

Source count: 8

Coverage: 50%

Last proof check: 2026-03-30T21:53:42.557Z

Signal Canvas receipt window

Watch and verify: SALMUBench: A Benchmark for Sensitive Association-Level Multimodal Unlearning

/buildability/salmubench-a-benchmark-for-sensitive-association-level-multimodal-unlearning

Watchwatch

Subject: SALMUBench: A Benchmark for Sensitive Association-Level Multimodal Unlearning

Verdict

Watch

Verdict is Watch because viability or proof quality is intermediate and should be re-evaluated before execution.

Preparing verified analysis

GitHub Code Pulse

No public code linked for this paper yet.

Claim map

Strong 12Mixed 0Weak 0

Evidencepartial
We introduce SALMUBench (Sensitive Association-Level Multimodal Unlearning), a benchmark built upon a synthetic dataset of 60K persona-attribute associations and two foundational models
Implicationpartial
The title and abstract explicitly state the purpose and name of the benchmark.
Verificationpartial
partial
Evidencepartial
a synthetic dataset of 60K persona-attribute associations
Implicationpartial
The abstract clearly states the size and nature of the dataset used for the benchmark.
Verificationpartial
partial
Evidencepartial
We propose a novel evaluation protocol with structured holdout sets (holdout identity, holdout association) to precisely measure unlearning efficacy and collateral damage.
Implicationpartial
The abstract describes the novel evaluation protocol and its components.
Verificationpartial
partial
Evidencepartial
current methods exhibit distinct failure modes: they either fail to forget effectively or over-generalize by erasing more than intended.
Implicationpartial
The abstract summarizes the findings of the benchmark regarding existing methods.
Verificationpartial
partial
Evidencepartial
our models achieve performance comparable to the original CLIP ViT-B/16 model [21] trained on 400M private pairs.
Implicationpartial
Table 1 and the accompanying text indicate that the 'Clean' and 'Compromised' models perform similarly to a public CLIP baseline.
Verificationpartial
partial
Evidencepartial
theCompromised model exhibits strong semantic alignment (memorization), while theCleanmodel yields significantly lower scores, consistent with t
Implicationpartial
Figure 5 and the text describe the difference in cosine similarity scores between the 'Compromised' and 'Clean' models on the sensitive set.
Verificationpartial
partial
Evidencepartial
Instead of evaluating for-getting via QA tasks in generative chat-based models, our benchmark directly evaluates forgetting capabilities within pretrained embedding spaces.
Implicationpartial
The abstract and introductory sections highlight this technical distinction of SALMUBench.
Verificationpartial
partial
Evidencepartial
We introduce SALMUBench (Sensitive Association-Level Multimodal Unlearning), a benchmark built upon a synthetic dataset of 60K persona-attribute associations and two foundational models: a Compromised model polluted with this data, and a Clean model without it.
Implicationpartial
The title and abstract explicitly state the purpose and name of the benchmark.
Verificationpartial
partial
Evidencepartial
We introduce SALMUBench (Sensitive Association-Level Multimodal Unlearning), a benchmark built upon a synthetic dataset of 60K persona-attribute associations and two foundational models: a Compromised model polluted with this data, and a Clean model without it.
Implicationpartial
The abstract clearly states the size and nature of the dataset used for the benchmark.
Verificationpartial
partial
Evidencepartial
We propose a novel evaluation protocol with structured holdout sets (holdout identity, holdout association) to precisely measure unlearning efficacy and collateral damage.
Implicationpartial
The abstract explicitly describes the novel evaluation protocol and its purpose.
Verificationpartial
partial
Evidencepartial
Our benchmark reveals that while utility-efficient deletion is feasible, current methods exhibit distinct failure modes: they either fail to forget effectively or over-generalize by erasing more than intended.
Implicationpartial
The abstract summarizes the findings of the benchmark regarding the performance of current methods.
Verificationpartial
partial
Evidencepartial
All models are trained from scratch on∼400M image-text pairs for 32 epochs.
Implicationpartial
Table 1 and the accompanying text indicate that the 'Clean' and 'Compromised' models trained on 400M pairs are comparable to a public CLIP baseline.
Verificationpartial
partial

Author intelligence and commercialization panels stay hidden until the proof receipt is verified, cites at least 3 references, includes at least 2 sources, and clears 50% coverage. The paper narrative and citation surfaces remain public while verification is pending.

SALMUBench: A Benchmark for Sensitive Association-Level Multimodal Unlearning

Use Signal Canvas as the narrative proof surface

Use this Signal Canvas via API or MCP

Signal Canvas proof surface