Upload folder using huggingface_hub

config.json

@@ -8,28 +8,20 @@
                 6
             ]
         },
-        "observation.images.camera1": {
+        "observation.images.arm": {
             "type": "VISUAL",
             "shape": [
                 3,
-                256,
-                256
+                240,
+                320
             ]
         },
-        "observation.images.camera2": {
+        "observation.images.above": {
             "type": "VISUAL",
             "shape": [
                 3,
-                256,
-                256
-            ]
-        },
-        "observation.images.camera3": {
-            "type": "VISUAL",
-            "shape": [
-                3,
-                256,
-                256
+                240,
+                320
             ]
         }
     },
@@ -45,11 +37,11 @@
     "use_amp": false,
     "use_peft": false,
     "push_to_hub": true,
-    "repo_id": "jailcode/smolvla_so101",
+    "repo_id": "jailcode/smolvla_base",
     "private": null,
     "tags": null,
     "license": null,
-    "pretrained_path": "lerobot/smolvla_base",
+    "pretrained_path": null,
     "chunk_size": 50,
     "n_action_steps": 50,
     "normalization_mapping": {
@@ -79,17 +71,17 @@
     ],
     "optimizer_eps": 1e-08,
     "optimizer_weight_decay": 1e-10,
-    "optimizer_grad_clip_norm": 10.0,
+    "optimizer_grad_clip_norm": 10,
     "scheduler_warmup_steps": 1000,
     "scheduler_decay_steps": 30000,
     "scheduler_decay_lr": 2.5e-06,
     "vlm_model_name": "HuggingFaceTB/SmolVLM2-500M-Video-Instruct",
-    "load_vlm_weights": true,
+    "load_vlm_weights": false,
     "add_image_special_tokens": false,
     "attention_mode": "cross_attn",
-    "prefix_length": 0,
-    "pad_language_to": "max_length",
-    "num_expert_layers": 0,
+    "prefix_length": -1,
+    "pad_language_to": "longest",
+    "num_expert_layers": -1,
     "num_vlm_layers": 16,
     "self_attn_every_n_layers": 2,
     "expert_width_multiplier": 0.75,

model.safetensors

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:119c51e0f542622b2646e15a44c6318003d9d1b8630a208d5034a6a57efab338
-size 906712520
+oid sha256:8e43c2f51aa6e5fce84e36acb6b6eeb72d22d9a41916d74b97ca8a96d266c879
+size 1197789224

policy_postprocessor_step_0_unnormalizer_processor.safetensors

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:d40b15f8d1c0d874ba40bd3a2221a30e6a8d1a957c34bb28bc5a190ea2d10aa0
+oid sha256:8b590da0b184eed90f9f24e2b9e605838baf03f926f457400ca531a8a5f6a3dd
 size 7536

policy_preprocessor.json

@@ -4,10 +4,7 @@
     {
       "registry_name": "rename_observations_processor",
       "config": {
-        "rename_map": {
-          "observation.images.top": "observation.images.camera1",
-          "observation.images.wrist": "observation.images.camera2"
-        }
+        "rename_map": {}
       }
     },
     {
@@ -24,7 +21,7 @@
         "max_length": 48,
         "task_key": "task",
         "padding_side": "right",
-        "padding": "max_length",
+        "padding": "longest",
         "truncation": true,
         "tokenizer_name": "HuggingFaceTB/SmolVLM2-500M-Video-Instruct"
       }
@@ -47,28 +44,20 @@
               6
             ]
           },
-          "observation.images.camera1": {
-            "type": "VISUAL",
-            "shape": [
-              3,
-              256,
-              256
-            ]
-          },
-          "observation.images.camera2": {
+          "observation.images.arm": {
             "type": "VISUAL",
             "shape": [
               3,
-              256,
-              256
+              240,
+              320
             ]
           },
-          "observation.images.camera3": {
+          "observation.images.above": {
             "type": "VISUAL",
             "shape": [
               3,
-              256,
-              256
+              240,
+              320
             ]
           },
           "action": {

policy_preprocessor_step_5_normalizer_processor.safetensors

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:d40b15f8d1c0d874ba40bd3a2221a30e6a8d1a957c34bb28bc5a190ea2d10aa0
+oid sha256:8b590da0b184eed90f9f24e2b9e605838baf03f926f457400ca531a8a5f6a3dd
 size 7536

train_config.json

@@ -1,6 +1,6 @@
 {
     "dataset": {
-        "repo_id": "jailcode/smolvla_50_episodes_1_20260604_121756",
+        "repo_id": "jailcode/data",
         "root": null,
         "episodes": null,
         "image_transforms": {
@@ -91,28 +91,20 @@
                     6
                 ]
             },
-            "observation.images.camera1": {
+            "observation.images.arm": {
                 "type": "VISUAL",
                 "shape": [
                     3,
-                    256,
-                    256
+                    240,
+                    320
                 ]
             },
-            "observation.images.camera2": {
+            "observation.images.above": {
                 "type": "VISUAL",
                 "shape": [
                     3,
-                    256,
-                    256
-                ]
-            },
-            "observation.images.camera3": {
-                "type": "VISUAL",
-                "shape": [
-                    3,
-                    256,
-                    256
+                    240,
+                    320
                 ]
             }
         },
@@ -128,11 +120,11 @@
         "use_amp": false,
         "use_peft": false,
         "push_to_hub": true,
-        "repo_id": "jailcode/smolvla_so101",
+        "repo_id": "jailcode/smolvla_base",
         "private": null,
         "tags": null,
         "license": null,
-        "pretrained_path": "lerobot/smolvla_base",
+        "pretrained_path": null,
         "chunk_size": 50,
         "n_action_steps": 50,
         "normalization_mapping": {
@@ -162,17 +154,17 @@
         ],
         "optimizer_eps": 1e-08,
         "optimizer_weight_decay": 1e-10,
-        "optimizer_grad_clip_norm": 10.0,
+        "optimizer_grad_clip_norm": 10,
         "scheduler_warmup_steps": 1000,
         "scheduler_decay_steps": 30000,
         "scheduler_decay_lr": 2.5e-06,
         "vlm_model_name": "HuggingFaceTB/SmolVLM2-500M-Video-Instruct",
-        "load_vlm_weights": true,
+        "load_vlm_weights": false,
         "add_image_special_tokens": false,
         "attention_mode": "cross_attn",
-        "prefix_length": 0,
-        "pad_language_to": "max_length",
-        "num_expert_layers": 0,
+        "prefix_length": -1,
+        "pad_language_to": "longest",
+        "num_expert_layers": -1,
         "num_vlm_layers": 16,
         "self_attn_every_n_layers": 2,
         "expert_width_multiplier": 0.75,
@@ -183,8 +175,8 @@
         "compile_mode": "max-autotune"
     },
     "reward_model": null,
-    "output_dir": "outputs/train/test_smolvla_3",
-    "job_name": "my_smolvla_so101",
+    "output_dir": "outputs/train/test_smolvla_4",
+    "job_name": "smolvla_so101_4",
     "resume": false,
     "seed": 1000,
     "cudnn_deterministic": false,
@@ -203,7 +195,7 @@
         "type": "adamw",
         "lr": 0.0001,
         "weight_decay": 1e-10,
-        "grad_clip_norm": 10.0,
+        "grad_clip_norm": 10,
         "betas": [
             0.9,
             0.95
@@ -234,9 +226,6 @@
     },
     "peft": null,
     "sample_weighting": null,
-    "rename_map": {
-        "observation.images.top": "observation.images.camera1",
-        "observation.images.wrist": "observation.images.camera2"
-    },
+    "rename_map": {},
     "checkpoint_path": null
 }