DiffusionPolicy for Pusht trained with keypoints of the current T position and the end T position

config.json

@@ -44,6 +44,13 @@
     "use_group_norm": true,
     "spatial_softmax_num_keypoints": 32,
     "use_separate_rgb_encoder_per_camera": false,
+    "state_backbone": "MLP",
+    "state_encoder_block_channels": [
+        64,
+        128
+    ],
+    "state_encoder_feature_dim": 256,
+    "state_encoder_use_layernorm": true,
     "down_dims": [
         512,
         1024,
@@ -61,8 +68,10 @@
     "prediction_type": "epsilon",
     "clip_sample": true,
     "clip_sample_range": 1.0,
+    "variance_type": "fixed_small",
     "num_inference_steps": null,
     "do_mask_loss_for_padding": false,
+    "optimizer_type": "adamw",
     "optimizer_lr": 0.0001,
     "optimizer_betas": [
         0.95,
@@ -71,12 +80,5 @@
     "optimizer_eps": 1e-08,
     "optimizer_weight_decay": 1e-06,
     "scheduler_name": "cosine",
-    "scheduler_warmup_steps": 500,
-    "state_backbone": "MLP",
-    "state_encoder_block_channels": [
-        64,
-        128
-    ],
-    "state_encoder_feature_dim": 256,
-    "state_encoder_use_layernorm": true
+    "scheduler_warmup_steps": 500
 }

model.safetensors

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:b08ae093c6ffc3eea669a042312db65cd45c6c2e1a0a40a24bec38d2c3f909ae
+oid sha256:472429110387bc740655c694dd71ed4467173cb346928977f66c89725523a461
 size 1050279112