Upload policy weights, train config and readme

README.md

@@ -1,13 +1,13 @@
 ---
-datasets: longzhiying/pick_cube2
+datasets: longzhiying/rokae_assembly_reward
 library_name: lerobot
 license: apache-2.0
 model_name: reward_classifier
 pipeline_tag: robotics
 tags:
-- reward_classifier
 - lerobot
 - robotics
+- reward_classifier
 ---
 
 # Model Card for reward_classifier

config.json

@@ -2,7 +2,7 @@
     "type": "reward_classifier",
     "n_obs_steps": 1,
     "input_features": {
-        "observation.images.top": {
+        "observation.images.side": {
             "type": "VISUAL",
             "shape": [
                 3,
@@ -10,7 +10,15 @@
                 128
             ]
         },
-        "observation.images.wrist": {
+        "observation.images.wrist1": {
+            "type": "VISUAL",
+            "shape": [
+                3,
+                128,
+                128
+            ]
+        },
+        "observation.images.wrist2": {
             "type": "VISUAL",
             "shape": [
                 3,
@@ -23,7 +31,7 @@
         "action": {
             "type": "ACTION",
             "shape": [
-                4
+                3
             ]
         }
     },
@@ -43,7 +51,7 @@
     "dropout_rate": 0.1,
     "model_name": "helper2424/resnet10",
     "model_type": "cnn",
-    "num_cameras": 2,
+    "num_cameras": 3,
     "learning_rate": 0.0001,
     "weight_decay": 0.01,
     "grad_clip_norm": 1.0,

model.safetensors

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:29688f959520ede5709da3bc1e8233131cb856ffd69d9acfe50cd2f1ad9e48ae
-size 29083940
+oid sha256:f05c42db8b55bb76f19be5cbf3ba456ecb9a73d3d1fab249753a230503bf1eb7
+size 33810260

train_config.json

@@ -1,6 +1,6 @@
 {
     "dataset": {
-        "repo_id": "longzhiying/pick_cube2",
+        "repo_id": "longzhiying/rokae_assembly_reward",
         "root": null,
         "episodes": null,
         "image_transforms": {
@@ -84,7 +84,7 @@
         "type": "reward_classifier",
         "n_obs_steps": 1,
         "input_features": {
-            "observation.images.top": {
+            "observation.images.side": {
                 "type": "VISUAL",
                 "shape": [
                     3,
@@ -92,7 +92,15 @@
                     128
                 ]
             },
-            "observation.images.wrist": {
+            "observation.images.wrist1": {
+                "type": "VISUAL",
+                "shape": [
+                    3,
+                    128,
+                    128
+                ]
+            },
+            "observation.images.wrist2": {
                 "type": "VISUAL",
                 "shape": [
                     3,
@@ -105,7 +113,7 @@
             "action": {
                 "type": "ACTION",
                 "shape": [
-                    4
+                    3
                 ]
             }
         },
@@ -125,7 +133,7 @@
         "dropout_rate": 0.1,
         "model_name": "helper2424/resnet10",
         "model_type": "cnn",
-        "num_cameras": 2,
+        "num_cameras": 3,
         "learning_rate": 0.0001,
         "weight_decay": 0.01,
         "grad_clip_norm": 1.0,
@@ -133,12 +141,12 @@
             "VISUAL": "MEAN_STD"
         }
     },
-    "output_dir": "outputs/train/2025-12-29/14-36-10_reward-classifier",
+    "output_dir": "outputs/train/2026-02-03/14-35-22_reward-classifier",
     "job_name": "reward-classifier",
     "resume": false,
     "seed": 2,
     "num_workers": 4,
-    "batch_size": 50,
+    "batch_size": 64,
     "steps": 5000,
     "eval_freq": 1000,
     "log_freq": 10,
@@ -168,7 +176,7 @@
         "project": "reward-classifier",
         "entity": null,
         "notes": null,
-        "run_id": "dfpmk9eg",
+        "run_id": "apcbu561",
         "mode": null
     },
     "checkpoint_path": null,