anonym371/dose-response-training-annotations

剂量响应训练数据安全注释数据集包含约870万张图像的安全注释（由LlavaGuard-7B和Gemini生成），按图像ID索引。不包含实际图像。数据来源包括lehduong/flux_generated、LucasFang/FLUX-Reason-6M和brivangl/midjourney-v6-llava三个数据集。The Dose-Response Training Data Safety Annotations dataset contains ~8.7M image safety annotations (LlavaGuard-7B + Gemini) indexed by image ID. No images included. Source datasets: lehduong/flux_generated, LucasFang/FLUX-Reason-6M, brivangl/midjourney-v6-llava.