Dates & evaluation details
VLSBench & MM-SafetyBench · Text-only · attack success · default evaluation · 2026-10-27
Published model comparison, 2023. MM-SafetyBench · default evaluation. Published average attack-success rate over thirteen scenario categories; lower is better. The named input condition and model size remain separate, and these values are not VLSBench safety accuracy. Table 3; source precision and any reported uncertainty retained. These are source-reported observations, not evaluations run by RLTasks.
VLSBench & MM-SafetyBench · SD · attack success · default evaluation · 2026-10-27
Published model comparison, 2023. MM-SafetyBench · default evaluation. Published average attack-success rate over thirteen scenario categories; lower is better. The named input condition and model size remain separate, and these values are not VLSBench safety accuracy. Table 3; source precision and any reported uncertainty retained. These are source-reported observations, not evaluations run by RLTasks.
VLSBench & MM-SafetyBench · Typo. · attack success · default evaluation · 2026-10-27
Published model comparison, 2023. MM-SafetyBench · default evaluation. Published average attack-success rate over thirteen scenario categories; lower is better. The named input condition and model size remain separate, and these values are not VLSBench safety accuracy. Table 3; source precision and any reported uncertainty retained. These are source-reported observations, not evaluations run by RLTasks.
VLSBench & MM-SafetyBench · SD+Typo. · attack success · default evaluation · 2026-10-27
Published model comparison, 2023. MM-SafetyBench · default evaluation. Published average attack-success rate over thirteen scenario categories; lower is better. The named input condition and model size remain separate, and these values are not VLSBench safety accuracy. Table 3; source precision and any reported uncertainty retained. These are source-reported observations, not evaluations run by RLTasks.
VLSBench & MM-SafetyBench · Text-only · attack success · default evaluation · 2026-10-27
Published model comparison, 2023. MM-SafetyBench · default evaluation. Published average attack-success rate over thirteen scenario categories; lower is better. The named input condition and model size remain separate, and these values are not VLSBench safety accuracy. Table 3; source precision and any reported uncertainty retained. These are source-reported observations, not evaluations run by RLTasks.
VLSBench & MM-SafetyBench · SD · attack success · default evaluation · 2026-10-27
Published model comparison, 2023. MM-SafetyBench · default evaluation. Published average attack-success rate over thirteen scenario categories; lower is better. The named input condition and model size remain separate, and these values are not VLSBench safety accuracy. Table 3; source precision and any reported uncertainty retained. These are source-reported observations, not evaluations run by RLTasks.
VLSBench & MM-SafetyBench · Typo. · attack success · default evaluation · 2026-10-27
Published model comparison, 2023. MM-SafetyBench · default evaluation. Published average attack-success rate over thirteen scenario categories; lower is better. The named input condition and model size remain separate, and these values are not VLSBench safety accuracy. Table 3; source precision and any reported uncertainty retained. These are source-reported observations, not evaluations run by RLTasks.
VLSBench & MM-SafetyBench · SD+Typo. · attack success · default evaluation · 2026-10-27
Published model comparison, 2023. MM-SafetyBench · default evaluation. Published average attack-success rate over thirteen scenario categories; lower is better. The named input condition and model size remain separate, and these values are not VLSBench safety accuracy. Table 3; source precision and any reported uncertainty retained. These are source-reported observations, not evaluations run by RLTasks.
VLSBench & MM-SafetyBench · Text-only · attack success · default evaluation · 2026-10-27
Published model comparison, 2023. MM-SafetyBench · default evaluation. Published average attack-success rate over thirteen scenario categories; lower is better. The named input condition and model size remain separate, and these values are not VLSBench safety accuracy. Table 6; source precision and any reported uncertainty retained. These are source-reported observations, not evaluations run by RLTasks.
VLSBench & MM-SafetyBench · SD · attack success · default evaluation · 2026-10-27
Published model comparison, 2023. MM-SafetyBench · default evaluation. Published average attack-success rate over thirteen scenario categories; lower is better. The named input condition and model size remain separate, and these values are not VLSBench safety accuracy. Table 6; source precision and any reported uncertainty retained. These are source-reported observations, not evaluations run by RLTasks.
VLSBench & MM-SafetyBench · Typo. · attack success · default evaluation · 2026-10-27
Published model comparison, 2023. MM-SafetyBench · default evaluation. Published average attack-success rate over thirteen scenario categories; lower is better. The named input condition and model size remain separate, and these values are not VLSBench safety accuracy. Table 6; source precision and any reported uncertainty retained. These are source-reported observations, not evaluations run by RLTasks.
VLSBench & MM-SafetyBench · SD+Typo. · attack success · default evaluation · 2026-10-27
Published model comparison, 2023. MM-SafetyBench · default evaluation. Published average attack-success rate over thirteen scenario categories; lower is better. The named input condition and model size remain separate, and these values are not VLSBench safety accuracy. Table 6; source precision and any reported uncertainty retained. These are source-reported observations, not evaluations run by RLTasks.