| { |
| "model_name": "HumanFace2ComicFaceGenerator", |
| "version": "6.0.0", |
| "description": "Progressive GAN for converting human faces to comic-style illustrations", |
| "author": "Your Name", |
| "license": "MIT", |
| "created_date": "2026-06-16", |
|
|
| "model": { |
| "generator": { |
| "type": "UNetEncoder-Decoder", |
| "input_nc": 3, |
| "output_nc": 3, |
| "ngf": 64, |
| "n_res_blocks": { |
| "64": 2, |
| "128": 4, |
| "256": 6 |
| }, |
| "dropout": 0.3, |
| "parameters": 35700000 |
| }, |
| "discriminator": { |
| "type": "PatchGAN", |
| "input_nc": 6, |
| "ndf": 64, |
| "n_layers": 3, |
| "parameters": 2770000 |
| } |
| }, |
|
|
| "training": { |
| "stages": [ |
| { |
| "stage": 1, |
| "resolution": 64, |
| "epochs": 50, |
| "batch_size": 8, |
| "gradient_accumulation": 1, |
| "final_val_loss": 22.14, |
| "final_disc_loss": 0.53 |
| }, |
| { |
| "stage": 2, |
| "resolution": 128, |
| "epochs": 30, |
| "batch_size": 4, |
| "gradient_accumulation": 1, |
| "final_val_loss": 20.16, |
| "final_disc_loss": 0.45 |
| }, |
| { |
| "stage": 3, |
| "resolution": 256, |
| "epochs": 20, |
| "batch_size": 2, |
| "gradient_accumulation": 4, |
| "final_val_loss": 17.83, |
| "final_disc_loss": 0.46 |
| } |
| ], |
| "optimizer": { |
| "type": "Adam", |
| "learning_rate": 0.0002, |
| "beta1": 0.5, |
| "beta2": 0.999, |
| "weight_decay": 0.0001 |
| }, |
| "loss_weights": { |
| "adversarial": 1.0, |
| "l1": 10, |
| "perceptual": 5, |
| "feature_matching": 2, |
| "frequency": 1 |
| }, |
| "data_augmentation": { |
| "type": "DiffAugment", |
| "operations": ["color", "translation", "cutout"], |
| "random_flip": true |
| }, |
| "regularization": { |
| "spectral_normalization": true, |
| "group_normalization": true, |
| "gradient_clipping": 1.0, |
| "ema_decay": 0.999, |
| "label_smoothing": { |
| "real_label": 0.9, |
| "fake_label": 0.1 |
| } |
| } |
| }, |
|
|
| "dataset": { |
| "name": "P2P-GAN Human2Comic Face Dataset", |
| "source": "https://www.kaggle.com/datasets/noob786/p2pgan-human2comic-face-paired-dataset", |
| "split": { |
| "train": 9900, |
| "validation": 100, |
| "test": 0 |
| }, |
| "image_size": [1024, 512], |
| "processed_size": [256, 256], |
| "description": "10,000 paired human and comic face images" |
| }, |
|
|
| "evaluation": { |
| "metrics": { |
| "inception_score": 5.2, |
| "fid_score": 45.3, |
| "inference_time_ms": 50 |
| }, |
| "hardware": { |
| "training": "NVIDIA A100", |
| "training_hours": 100, |
| "inference_memory_vram_gb": 6, |
| "inference_memory_ram_gb": 8 |
| } |
| }, |
|
|
| "inference": { |
| "framework": "PyTorch", |
| "device": "cuda", |
| "precision": "float32", |
| "batch_size": 1, |
| "input_format": "PIL.Image or numpy.ndarray", |
| "output_format": "PIL.Image" |
| }, |
|
|
| "deployment": { |
| "platforms": [ |
| "HuggingFace Spaces", |
| "Local (GPU/CPU)", |
| "Docker", |
| "API" |
| ], |
| "api_framework": "Gradio", |
| "api_version": "6.18.0", |
| "port": 7860, |
| "dependencies": [ |
| "torch>=2.0.0", |
| "torchvision>=0.15.0", |
| "transformers>=4.30.0", |
| "gradio>=6.0.0", |
| "pillow>=9.0.0", |
| "numpy>=1.23.0" |
| ] |
| }, |
|
|
| "capabilities": { |
| "image_to_image": true, |
| "batch_processing": false, |
| "real_time_video": false, |
| "gpu_accelerated": true, |
| "quantized_variant": false, |
| "multi_resolution": true, |
| "style_transfer": true |
| }, |
|
|
| "limitations": { |
| "requires_gpu": false, |
| "best_on_gpu": true, |
| "works_on_cpu": true, |
| "requires_faces": true, |
| "single_face_per_image": true, |
| "requires_frontality": "recommended", |
| "min_face_size_pixels": 64 |
| }, |
|
|
| "checkpoints": { |
| "generator_best": { |
| "file": "./model_weights/generator_best.pth", |
| "stage": 3, |
| "resolution": 256, |
| "size_mb": 246 |
| } |
| } |
| } |
|
|
|
|
