Upload config.json with huggingface_hub
Browse files- config.json +18 -6
config.json
CHANGED
|
@@ -1,5 +1,5 @@
|
|
| 1 |
{
|
| 2 |
-
"save_path": "
|
| 3 |
"model": {
|
| 4 |
"base_learning_rate": 8e-05,
|
| 5 |
"target": "latent_diffusion.ldm.models.diffusion.ddpm.LatentDiffusion",
|
|
@@ -28,7 +28,7 @@
|
|
| 28 |
64,
|
| 29 |
48
|
| 30 |
],
|
| 31 |
-
"in_channels":
|
| 32 |
"out_channels": 4,
|
| 33 |
"model_channels": 192,
|
| 34 |
"attention_resolutions": [
|
|
@@ -48,6 +48,18 @@
|
|
| 48 |
"transformer_depth": 1
|
| 49 |
}
|
| 50 |
},
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 51 |
"first_stage_config": {
|
| 52 |
"target": "latent_diffusion.ldm.models.autoencoder.AutoencoderKL",
|
| 53 |
"params": {
|
|
@@ -81,8 +93,8 @@
|
|
| 81 |
"data": {
|
| 82 |
"target": "data.data_processing.datasets.DataModule",
|
| 83 |
"params": {
|
| 84 |
-
"batch_size":
|
| 85 |
-
"num_workers":
|
| 86 |
"wrap": false,
|
| 87 |
"shuffle": true,
|
| 88 |
"drop_last": true,
|
|
@@ -92,7 +104,7 @@
|
|
| 92 |
"train": {
|
| 93 |
"target": "data.data_processing.datasets.ActionsData",
|
| 94 |
"params": {
|
| 95 |
-
"data_csv_path": "
|
| 96 |
"normalization": "standard",
|
| 97 |
"context_length": 32
|
| 98 |
}
|
|
@@ -106,7 +118,7 @@
|
|
| 106 |
"limit_val_batches": 0,
|
| 107 |
"accelerator": "gpu",
|
| 108 |
"gpus": 1,
|
| 109 |
-
"accumulate_grad_batches":
|
| 110 |
"gradient_clip_val": 1,
|
| 111 |
"checkpoint_callback": true
|
| 112 |
}
|
|
|
|
| 1 |
{
|
| 2 |
+
"save_path": "saved_standard_challenging_context32_nocond_cont_cont_all_cont_eval",
|
| 3 |
"model": {
|
| 4 |
"base_learning_rate": 8e-05,
|
| 5 |
"target": "latent_diffusion.ldm.models.diffusion.ddpm.LatentDiffusion",
|
|
|
|
| 28 |
64,
|
| 29 |
48
|
| 30 |
],
|
| 31 |
+
"in_channels": 8,
|
| 32 |
"out_channels": 4,
|
| 33 |
"model_channels": 192,
|
| 34 |
"attention_resolutions": [
|
|
|
|
| 48 |
"transformer_depth": 1
|
| 49 |
}
|
| 50 |
},
|
| 51 |
+
"temporal_encoder_config": {
|
| 52 |
+
"target": "latent_diffusion.ldm.modules.encoders.temporal_encoder.TemporalEncoder",
|
| 53 |
+
"params": {
|
| 54 |
+
"input_channels": 6,
|
| 55 |
+
"hidden_size": 1024,
|
| 56 |
+
"num_layers": 1,
|
| 57 |
+
"dropout": 0.1,
|
| 58 |
+
"output_channels": 4,
|
| 59 |
+
"output_height": 48,
|
| 60 |
+
"output_width": 64
|
| 61 |
+
}
|
| 62 |
+
},
|
| 63 |
"first_stage_config": {
|
| 64 |
"target": "latent_diffusion.ldm.models.autoencoder.AutoencoderKL",
|
| 65 |
"params": {
|
|
|
|
| 93 |
"data": {
|
| 94 |
"target": "data.data_processing.datasets.DataModule",
|
| 95 |
"params": {
|
| 96 |
+
"batch_size": 8,
|
| 97 |
+
"num_workers": 1,
|
| 98 |
"wrap": false,
|
| 99 |
"shuffle": true,
|
| 100 |
"drop_last": true,
|
|
|
|
| 104 |
"train": {
|
| 105 |
"target": "data.data_processing.datasets.ActionsData",
|
| 106 |
"params": {
|
| 107 |
+
"data_csv_path": "desktop_sequences_filtered_with_desktop_1.5k.challenging.train.target_frames.csv",
|
| 108 |
"normalization": "standard",
|
| 109 |
"context_length": 32
|
| 110 |
}
|
|
|
|
| 118 |
"limit_val_batches": 0,
|
| 119 |
"accelerator": "gpu",
|
| 120 |
"gpus": 1,
|
| 121 |
+
"accumulate_grad_batches": 999999,
|
| 122 |
"gradient_clip_val": 1,
|
| 123 |
"checkpoint_callback": true
|
| 124 |
}
|