Upload policy weights, train config and readme

README.md

@@ -6,9 +6,9 @@ license: apache-2.0
 model_name: smolvla
 pipeline_tag: robotics
 tags:
-- robotics
 - smolvla
 - lerobot
+- robotics
 ---
 
 # Model Card for smolvla
@@ -85,7 +85,7 @@ The policy consumes these observation features and produces these action feature
 
 | Setting | Value |
 | --- | --- |
-| Training steps | 2 |
+| Training steps | 200 |
 | Batch size | 256 |
 | Optimizer | adamw |
 | Learning rate | 0.0001 |

model.safetensors

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:59be5e7cd5e1fd13bf9320b2b206729de5a155d524aa615271ac9027e06b39cd
+oid sha256:6899ce699e7be5f8bd9efa5e05b31effd0a732459e2ef75ffb90f83231ee27eb
 size 906712520

train_config.json

@@ -78,7 +78,7 @@
         "revision": null,
         "use_imagenet_stats": true,
         "video_backend": "torchcodec",
-        "return_uint8": false,
+        "return_uint8": true,
         "depth_output_unit": "mm",
         "streaming": false,
         "eval_split": 0.0
@@ -187,8 +187,8 @@
         "compile_mode": "default"
     },
     "reward_model": null,
-    "output_dir": "/work/sim_vla_ws/runs/train/smolvla_run_post_fix",
-    "job_name": "smolvla_run_post_fix",
+    "output_dir": "/work/sim_vla_ws/runs/train/smolvla_ret_u8",
+    "job_name": "smolvla_ret_u8",
     "resume": false,
     "seed": 0,
     "cudnn_deterministic": false,
@@ -197,9 +197,9 @@
     "prefetch_factor": 4,
     "persistent_workers": true,
     "dataloader_multiprocessing_context": "spawn",
-    "steps": 4000,
+    "steps": 200,
     "env_eval_freq": 20000,
-    "log_freq": 200,
+    "log_freq": 50,
     "eval_steps": 0,
     "max_eval_samples": 0,
     "tolerance_s": 0.0001,
@@ -238,7 +238,7 @@
         "project": "lerobot",
         "entity": null,
         "notes": null,
-        "run_id": "vostlw4n",
+        "run_id": "7vlt74wy",
         "mode": null,
         "add_tags": true
     },