| { |
| "library_name": "zeromodels", |
| "zeromodels_version": "1.3.3", |
| "model_module": "zeromodels.diffusers", |
| "model_class": "ZMDiffusion", |
| "variant": "stable-diffusion-3-medium", |
| "weights": "model.weights.json", |
| "schema_version": 2, |
| "weight_dtype": "float16", |
| "weight_dtype_overrides": { |
| "vae_encoder": "float32", |
| "vae_decoder": "float32" |
| }, |
| "model_type": "stable_diffusion_3", |
| "transformer_config": { |
| "sample_size": 128, |
| "patch_size": 2, |
| "in_channels": 16, |
| "out_channels": 16, |
| "num_layers": 24, |
| "attention_head_dim": 64, |
| "num_attention_heads": 24, |
| "joint_attention_dim": 4096, |
| "caption_projection_dim": 1536, |
| "pooled_projection_dim": 2048, |
| "pos_embed_max_size": 192, |
| "qk_norm": null, |
| "dual_attention_layers": [], |
| "text_seq_len": 333 |
| }, |
| "vae_config": { |
| "in_channels": 3, |
| "out_channels": 3, |
| "latent_channels": 16, |
| "block_out_channels": [ |
| 128, |
| 256, |
| 512, |
| 512 |
| ], |
| "layers_per_block": 2, |
| "norm_num_groups": 32, |
| "sample_size": 1024, |
| "scaling_factor": 1.5305, |
| "force_upcast": true, |
| "shift_factor": 0.0609, |
| "use_quant_conv": false, |
| "use_post_quant_conv": false |
| }, |
| "text_config": { |
| "hidden_dim": 768, |
| "num_heads": 12, |
| "num_layers": 12, |
| "mlp_ratio": 4.0, |
| "vocab_size": 49408, |
| "max_seq_len": 77, |
| "projection_dim": 768, |
| "hidden_act": "quick_gelu" |
| }, |
| "text_config_2": { |
| "hidden_dim": 1280, |
| "num_heads": 20, |
| "num_layers": 32, |
| "mlp_ratio": 4.0, |
| "vocab_size": 49408, |
| "max_seq_len": 77, |
| "projection_dim": 1280, |
| "hidden_act": "gelu" |
| }, |
| "hidden_act": "quick_gelu", |
| "layer_norm_eps": 1e-05, |
| "bos_token_id": 49406, |
| "eos_token_id": 49407, |
| "pad_token_id": 49407, |
| "pad_token_id_2": 0, |
| "max_sequence_length": 256, |
| "eos_token_id_3": 1, |
| "pad_token_id_3": 0, |
| "scheduler_config": { |
| "_class_name": "FlowMatchEulerDiscreteScheduler", |
| "num_train_timesteps": 1000, |
| "shift": 3.0 |
| }, |
| "generate_args": { |
| "num_inference_steps": 28, |
| "guidance_scale": 7.0 |
| } |
| } |
|
|
