Upload policy weights, train config and readme

README.md

@@ -5,9 +5,9 @@ license: apache-2.0
 model_name: diffusion
 pipeline_tag: robotics
 tags:
-- robotics
 - diffusion
 - lerobot
+- robotics
 ---
 
 # Model Card for diffusion

config.json

@@ -2,7 +2,7 @@
     "type": "diffusion",
     "n_obs_steps": 1,
     "normalization_mapping": {
-        "VISUAL": "MEAN_STD",
+        "VISUAL": "IDENTITY",
         "STATE": "MIN_MAX",
         "ACTION": "MIN_MAX"
     },
@@ -39,20 +39,22 @@
     "license": null,
     "horizon": 64,
     "n_action_steps": 32,
+    "per_feature_normalization_overrides": {
+        "observation.images.depth": "MEAN_STD"
+    },
     "drop_n_last_frames": 7,
     "vision_backbone": "resnet18",
-    "crop_shape": [
-        224,
-        224
-    ],
+    "crop_shape": null,
     "crop_is_random": true,
     "pretrained_backbone_weights": null,
     "use_group_norm": true,
     "spatial_softmax_num_keypoints": 32,
-    "use_separate_rgb_encoder_per_camera": true,
+    "use_separate_rgb_encoder_per_camera": false,
     "use_depth_as_channel": true,
     "use_depth_only": true,
     "rgbd_pretrained_backbone_weights": null,
+    "norm_rgb": false,
+    "norm_depth": false,
     "down_dims": [
         512,
         1024,

model.safetensors

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:9e23c4927dc30ba827e2b4b220cf70a28ad1c3330a50c8b02b160cadaf7cbf3c
-size 1046065148
+oid sha256:a90a629fa734a04113be09559fef54be1487c454bc251b5b3202675e42f50183
+size 1046065348

train_config.json

@@ -69,7 +69,7 @@
         "type": "diffusion",
         "n_obs_steps": 1,
         "normalization_mapping": {
-            "VISUAL": "MEAN_STD",
+            "VISUAL": "IDENTITY",
             "STATE": "MIN_MAX",
             "ACTION": "MIN_MAX"
         },
@@ -106,20 +106,22 @@
         "license": null,
         "horizon": 64,
         "n_action_steps": 32,
+        "per_feature_normalization_overrides": {
+            "observation.images.depth": "MEAN_STD"
+        },
         "drop_n_last_frames": 7,
         "vision_backbone": "resnet18",
-        "crop_shape": [
-            224,
-            224
-        ],
+        "crop_shape": null,
         "crop_is_random": true,
         "pretrained_backbone_weights": null,
         "use_group_norm": true,
         "spatial_softmax_num_keypoints": 32,
-        "use_separate_rgb_encoder_per_camera": true,
+        "use_separate_rgb_encoder_per_camera": false,
         "use_depth_as_channel": true,
         "use_depth_only": true,
         "rgbd_pretrained_backbone_weights": null,
+        "norm_rgb": false,
+        "norm_depth": false,
         "down_dims": [
             512,
             1024,
@@ -149,7 +151,7 @@
         "scheduler_name": "cosine",
         "scheduler_warmup_steps": 500
     },
-    "output_dir": "outputs/train/2025-10-15/09-37-57_diffusion",
+    "output_dir": "outputs/train/2025-10-16/16-49-40_diffusion",
     "job_name": "diffusion",
     "resume": false,
     "seed": 1000,
@@ -184,11 +186,11 @@
     },
     "wandb": {
         "enable": true,
-        "disable_artifact": false,
-        "project": "lerobot",
+        "disable_artifact": true,
+        "project": "digit3a_rgbd",
         "entity": null,
         "notes": null,
-        "run_id": "g71jjm90",
+        "run_id": "egik8o1l",
         "mode": null
     }
 }