Spaces:
Sleeping
Sleeping
added config
Browse files
models/2024-03-01T23-15-36-project.yaml
ADDED
|
@@ -0,0 +1,90 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
model:
|
| 2 |
+
base_learning_rate: 5.0e-06
|
| 3 |
+
target: ldm.models.diffusion.ddpm.LatentDiffusion
|
| 4 |
+
project: Fish-Diffusion
|
| 5 |
+
params:
|
| 6 |
+
linear_start: 0.0015
|
| 7 |
+
linear_end: 0.0195
|
| 8 |
+
num_timesteps_cond: 1
|
| 9 |
+
log_every_t: 200
|
| 10 |
+
timesteps: 1000
|
| 11 |
+
first_stage_key: image
|
| 12 |
+
cond_stage_key: class_label
|
| 13 |
+
image_size: 64
|
| 14 |
+
channels: 3
|
| 15 |
+
cond_stage_trainable: true
|
| 16 |
+
conditioning_key: crossattn
|
| 17 |
+
monitor: val/loss_simple_ema
|
| 18 |
+
unet_config:
|
| 19 |
+
target: ldm.modules.diffusionmodules.openaimodel.UNetModel
|
| 20 |
+
params:
|
| 21 |
+
image_size: 64
|
| 22 |
+
in_channels: 3
|
| 23 |
+
out_channels: 3
|
| 24 |
+
model_channels: 224
|
| 25 |
+
attention_resolutions:
|
| 26 |
+
- 8
|
| 27 |
+
- 4
|
| 28 |
+
- 2
|
| 29 |
+
num_res_blocks: 2
|
| 30 |
+
channel_mult:
|
| 31 |
+
- 1
|
| 32 |
+
- 2
|
| 33 |
+
- 3
|
| 34 |
+
- 4
|
| 35 |
+
num_head_channels: 32
|
| 36 |
+
use_spatial_transformer: true
|
| 37 |
+
transformer_depth: 1
|
| 38 |
+
context_dim: 512
|
| 39 |
+
first_stage_config:
|
| 40 |
+
target: ldm.models.autoencoder.VQModelInterface
|
| 41 |
+
params:
|
| 42 |
+
embed_dim: 3
|
| 43 |
+
n_embed: 8192
|
| 44 |
+
ckpt_path: /fastscratch/mridul/new_diffusion_models/ldm/vq-base/2023-11-07T15-07-37_VQ_f4_org/checkpoints/epoch=000371.ckpt
|
| 45 |
+
ddconfig:
|
| 46 |
+
double_z: false
|
| 47 |
+
z_channels: 3
|
| 48 |
+
resolution: 256
|
| 49 |
+
in_channels: 3
|
| 50 |
+
out_ch: 3
|
| 51 |
+
ch: 128
|
| 52 |
+
ch_mult:
|
| 53 |
+
- 1
|
| 54 |
+
- 2
|
| 55 |
+
- 4
|
| 56 |
+
num_res_blocks: 2
|
| 57 |
+
attn_resolutions: []
|
| 58 |
+
dropout: 0.0
|
| 59 |
+
lossconfig:
|
| 60 |
+
target: torch.nn.Identity
|
| 61 |
+
cond_stage_config:
|
| 62 |
+
target: ldm.modules.encoders.modules.HeirClassEmbedder
|
| 63 |
+
params:
|
| 64 |
+
n_classes:
|
| 65 |
+
- 3
|
| 66 |
+
- 6
|
| 67 |
+
- 9
|
| 68 |
+
- 38
|
| 69 |
+
embed_dim: 512
|
| 70 |
+
key: class_to_node
|
| 71 |
+
data:
|
| 72 |
+
target: main.DataModuleFromConfig
|
| 73 |
+
params:
|
| 74 |
+
batch_size: 8
|
| 75 |
+
num_workers: 8
|
| 76 |
+
wrap: true
|
| 77 |
+
train:
|
| 78 |
+
target: ldm.data.custom.CustomTrain
|
| 79 |
+
params:
|
| 80 |
+
training_images_list_file: /fastscratch/elhamod/data/Fish/fish_train.txt
|
| 81 |
+
size: 256
|
| 82 |
+
add_labels: true
|
| 83 |
+
class_to_node: /fastscratch/mridul/fishes/class_to_ancestral_label.pkl
|
| 84 |
+
validation:
|
| 85 |
+
target: ldm.data.custom.CustomTest
|
| 86 |
+
params:
|
| 87 |
+
test_images_list_file: /fastscratch/elhamod/data/Fish/fish_test.txt
|
| 88 |
+
size: 256
|
| 89 |
+
add_labels: true
|
| 90 |
+
class_to_node: /fastscratch/mridul/fishes/class_to_ancestral_label.pkl
|