Upload config.json with huggingface_hub
Browse files- config.json +51 -0
config.json
ADDED
|
@@ -0,0 +1,51 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"architectures": ["FMNISTT2IModel"],
|
| 3 |
+
"model_type": "fmnist_t2i_diffusion",
|
| 4 |
+
"patch_size": 1,
|
| 5 |
+
"image_size": 8,
|
| 6 |
+
"num_channels": 4,
|
| 7 |
+
"embed_dim": 384,
|
| 8 |
+
"mlp_multiplier": 4,
|
| 9 |
+
"n_layers": 12,
|
| 10 |
+
"n_heads": 6,
|
| 11 |
+
"cross_heads": 4,
|
| 12 |
+
"noise_embed_dims": 128,
|
| 13 |
+
"text_emb_size": 10,
|
| 14 |
+
"dropout_prob": 0.1,
|
| 15 |
+
"use_flash_attention": true,
|
| 16 |
+
"text_encoding_type": "one_hot",
|
| 17 |
+
"text_dropout_prob": 0.1,
|
| 18 |
+
"num_classes": 10,
|
| 19 |
+
"text_embed_dim": 384,
|
| 20 |
+
"text_projection_layers": 1,
|
| 21 |
+
"diffusion_config": {
|
| 22 |
+
"timesteps": "continuous",
|
| 23 |
+
"beta_a": 1.0,
|
| 24 |
+
"beta_b": 2.5,
|
| 25 |
+
"noise_start": 0.99,
|
| 26 |
+
"noise_end": 0.01
|
| 27 |
+
},
|
| 28 |
+
"vae_config": {
|
| 29 |
+
"vae_path": "stabilityai/sd-vae-ft-mse",
|
| 30 |
+
"vae_latent_scale": 0.18215
|
| 31 |
+
},
|
| 32 |
+
"training_config": {
|
| 33 |
+
"lr": 0.001,
|
| 34 |
+
"weight_decay": 0.01,
|
| 35 |
+
"betas": [0.9, 0.999],
|
| 36 |
+
"eps": 1e-8,
|
| 37 |
+
"max_grad_norm": 1.0,
|
| 38 |
+
"ema_decay": 0.999
|
| 39 |
+
},
|
| 40 |
+
"hardware_config": {
|
| 41 |
+
"use_amp": true,
|
| 42 |
+
"amp_dtype": "float16",
|
| 43 |
+
"compile_model": false
|
| 44 |
+
},
|
| 45 |
+
"init_cfg": {
|
| 46 |
+
"type": "trunc_normal",
|
| 47 |
+
"std": 0.02,
|
| 48 |
+
"a": -2,
|
| 49 |
+
"b": 2
|
| 50 |
+
}
|
| 51 |
+
}
|