nsfw image detector This model is a fine tuned version of google/vit base patch16 224 in21k on the imagefolder dataset. It achieves the following results on the evaluation set: Loss: 0.8138 Accuracy: 0.9316 Accuracy K: 0.9887 You can access 384 version on: https://huggingface.co/LukeJacob2023/nsfw image detector 384 Model description Labels: ['drawings', 'hentai', 'neutral', 'porn', 'sexy'] Intended uses & limitations Training and evaluation data A custom dataset about 28k images, if you need to improve your domain's accurate, you can contribute the dataset to me. Training procedure Training hyperparameters The following hyperparameters were used during training: learning rate: 2e 05 train batch size: 32 eval batch size: 32 seed: 42 optimizer: Adam with betas=(0.9,0.999) and epsilon=1e 08 lr scheduler type: linear lr scheduler warmup steps: 500 num epochs: 10 mixed precision training: Native AMP Training results Training Loss Epoch Step Validation Loss Accuracy Accuracy K : : : : : : : : : : : : 0.7836 1.0 720 0.3188 0.9085 0.9891 0.2441 2.0 1440 0.2382 0.9257 0.9936 0.1412 3.0 2160 0.2334 0.9335 0.9932 0.0857 4.0 2880 0.2934 0.9347 0.9934 0.0569 5.0 3600 0.4500 0.9307 0.9927 0.037…
We use cookies for essential functionality and analytics. You can accept or reject analytics cookies.Cookie policy