update train-diffusion.json

Signed-off-by: Can-Zhao <[email protected]>

models/brats_mri_generative_diffusion/configs/train_diffusion.json

@@ -1,55 +1,42 @@
 {
     "ckpt_dir": "$@bundle_root + '/models'",
     "train_batch_size": 4,
-    "lr": 1e-05,
     "train_patch_size": [
-        144,
-        176,
-        112
+        192,
+        192,
+        128
     ],
     "latent_shape": [
         "@latent_channels",
-        36,
-        44,
-        28
+        48,
+        48,
+        32
     ],
     "load_autoencoder_path": "$@bundle_root + '/models/model_autoencoder.pt'",
     "load_autoencoder": "$@autoencoder_def.load_state_dict(torch.load(@load_autoencoder_path))",
     "autoencoder": "$@autoencoder_def.to(@device)",
-    "network_def": {
+    "diffusion_def": {
         "_target_": "generative.networks.nets.DiffusionModelUNet",
         "spatial_dims": "@spatial_dims",
         "in_channels": "@latent_channels",
         "out_channels": "@latent_channels",
-        "num_channels": [
-            256,
-            256,
-            512
-        ],
-        "attention_levels": [
-            false,
-            true,
-            true
-        ],
-        "num_head_channels": [
-            0,
-            64,
-            64
-        ],
-        "num_res_blocks": 2
+        "num_channels":[128, 256, 512],
+        "attention_levels":[false, true, true],
+        "num_head_channels":[0, 32, 32],
+        "num_res_blocks": 2,
+        "use_flash_attention": true
     },
-    "diffusion": "$@network_def.to(@device)",
+    "diffusion": "$@diffusion_def.to(@device)",
     "optimizer": {
         "_target_": "torch.optim.Adam",
         "params": "[email protected]()",
-        "lr": "@lr"
+        "lr": 1e-04
     },
     "lr_scheduler": {
         "_target_": "torch.optim.lr_scheduler.MultiStepLR",
         "optimizer": "@optimizer",
         "milestones": [
-            100,
-            1000
+            5000
         ],
         "gamma": 0.1
     },
@@ -59,20 +46,20 @@
         "_requires_": [
             "@load_autoencoder"
         ],
-        "schedule": "scaled_linear_beta",
+        "schedule": "linear_beta",
         "num_train_timesteps": 1000,
         "beta_start": 0.0015,
         "beta_end": 0.0195
     },
+    "inferer": {
+        "_target_": "generative.inferers.LatentDiffusionInferer",
+        "scheduler": "@noise_scheduler",
+        "scale_factor": "@scale_factor"
+    },
     "loss": {
         "_target_": "torch.nn.MSELoss"
     },
     "train": {
-        "inferer": {
-            "_target_": "generative.inferers.LatentDiffusionInferer",
-            "scheduler": "@noise_scheduler",
-            "scale_factor": "@scale_factor"
-        },
         "crop_transforms": [
             {
                 "_target_": "CenterSpatialCropd",
@@ -85,13 +72,8 @@
             "transforms": "$@preprocessing_transforms + @train#crop_transforms + @final_transforms"
         },
         "dataset": {
-            "_target_": "monai.apps.DecathlonDataset",
-            "root_dir": "@dataset_dir",
-            "task": "Task01_BrainTumour",
-            "section": "training",
-            "cache_rate": 1.0,
-            "num_workers": 8,
-            "download": false,
+            "_target_": "Dataset",
+            "data": "@train_datalist",
             "transform": "@train#preprocessing"
         },
         "dataloader": {
@@ -116,7 +98,7 @@
                 "save_interval": 0,
                 "save_final": true,
                 "epoch_level": true,
-                "final_filename": "model.pt"
+                "final_filename": "model_ldm.pt"
             },
             {
                 "_target_": "StatsHandler",
@@ -127,20 +109,21 @@
                 "_target_": "TensorBoardStatsHandler",
                 "log_dir": "@tf_dir",
                 "tag_name": "train_diffusion_loss",
-                "output_transform": "$lambda x: monai.handlers.from_engine(['loss'], first=True)(x)"
+                "output_transform": "$lambda x: monai.handlers.from_engine(['loss'], first=True)(x)",
+                "iteration_log":false
             }
         ],
         "trainer": {
             "_target_": "scripts.ldm_trainer.LDMTrainer",
             "device": "@device",
-            "max_epochs": 5000,
+            "max_epochs": 10000,
             "train_data_loader": "@train#dataloader",
             "network": "@diffusion",
             "autoencoder_model": "@autoencoder",
             "optimizer": "@optimizer",
             "loss_function": "@loss",
             "latent_shape": "@latent_shape",
-            "inferer": "@train#inferer",
+            "inferer": "@inferer",
             "key_train_metric": "$None",
             "train_handlers": "@train#handlers"
         }