Text2VLM: Adapting Text-Only Datasets to Evaluate Alignment Training in Visual Language Models

Name: Text2VLM: Adapting Text-Only Datasets to Evaluate Alignment Training in Visual Language Models
Uploaded: 2025-01-01
Duration: 19 min 43 s
Description: Text2VLM: Adapting Text-Only Datasets to Evaluate Alignment Training in Visual Language Models — a 2025 CAMLIS 2025 security conference talk on ai-ml (19:43 min). Watch the full video, indexed free on the Breach Arc Archive.

CAMLIS 2025 · 2025 · 19:43

▶

Text2VLM: Adapting Text-Only Datasets to Evaluate Alignment Training in Visual Language Models — a 2025 CAMLIS 2025 security conference talk on ai-ml (19:43 min). Watch the full video, indexed free on the Breach Arc Archive.

▶ Watch video on YouTube

ai-ml

← Back to the Breach Arc Archive