keras/stable_diffusion_3.5_large
012
1{2 "module": "keras_hub.src.models.stable_diffusion_3.stable_diffusion_3_backbone",3 "class_name": "StableDiffusion3Backbone",4 "config": {5 "name": "stable_diffusion_3.5_large_backbone",6 "trainable": true,7 "dtype": {8 "module": "keras",9 "class_name": "DTypePolicy",10 "config": {11 "name": "bfloat16"12 },13 "registered_name": null14 },15 "mmdit_patch_size": 2,16 "mmdit_hidden_dim": 2432,17 "mmdit_num_layers": 38,18 "mmdit_num_heads": 38,19 "mmdit_position_size": 192,20 "mmdit_qk_norm": "rms_norm",21 "mmdit_dual_attention_indices": null,22 "vae": {23 "module": "keras_hub.src.models.vae.vae_backbone",24 "class_name": "VAEBackbone",25 "config": {26 "name": "vae",27 "trainable": true,28 "dtype": {29 "module": "keras",30 "class_name": "DTypePolicy",31 "config": {32 "name": "bfloat16"33 },34 "registered_name": null35 },36 "encoder_num_filters": [37 128,38 256,39 512,40 51241 ],42 "encoder_num_blocks": [43 2,44 2,45 2,46 247 ],48 "decoder_num_filters": [49 512,50 512,51 256,52 12853 ],54 "decoder_num_blocks": [55 3,56 3,57 3,58 359 ],60 "sampler_method": "sample",61 "input_channels": 3,62 "sample_channels": 32,63 "output_channels": 3,64 "scale": 1.5305,65 "shift": 0.060966 },67 "registered_name": "VAEBackbone"68 },69 "clip_l": {70 "module": "keras_hub.src.models.clip.clip_text_encoder",71 "class_name": "CLIPTextEncoder",72 "config": {73 "name": "clip_l",74 "trainable": true,75 "dtype": {76 "module": "keras",77 "class_name": "DTypePolicy",78 "config": {79 "name": "float16"80 },81 "registered_name": null82 },83 "vocabulary_size": 49408,84 "embedding_dim": 768,85 "hidden_dim": 768,86 "num_layers": 12,87 "num_heads": 12,88 "intermediate_dim": 3072,89 "intermediate_activation": "quick_gelu",90 "intermediate_output_index": 10,91 "max_sequence_length": 7792 },93 "registered_name": "keras_hub>CLIPTextEncoder"94 },95 "clip_g": {96 "module": "keras_hub.src.models.clip.clip_text_encoder",97 "class_name": "CLIPTextEncoder",98 "config": {99 "name": "clip_g",100 "trainable": true,101 "dtype": {102 "module": "keras",103 "class_name": "DTypePolicy",104 "config": {105 "name": "float16"106 },107 "registered_name": null108 },109 "vocabulary_size": 49408,110 "embedding_dim": 1280,111 "hidden_dim": 1280,112 "num_layers": 32,113 "num_heads": 20,114 "intermediate_dim": 5120,115 "intermediate_activation": "gelu",116 "intermediate_output_index": 30,117 "max_sequence_length": 77118 },119 "registered_name": "keras_hub>CLIPTextEncoder"120 },121 "t5": null,122 "latent_channels": 16,123 "output_channels": 3,124 "num_train_timesteps": 1000,125 "shift": 3.0,126 "image_shape": [127 1024,128 1024,129 3130 ]131 },132 "registered_name": "keras_hub>StableDiffusion3Backbone"133}