Upload policy weights, train config and readme

README.md

@@ -1,22 +1,21 @@
 ---
-base_model: lerobot/smolvla_base
-datasets: aloha/push_T_top_to_bottom_mixture_5_100percent
+datasets: aloha/simulation_push_T_bottom_right
 library_name: lerobot
 license: apache-2.0
-model_name: smolvla
+model_name: act
 pipeline_tag: robotics
 tags:
-- lerobot
+- act
 - robotics
-- smolvla
+- lerobot
 ---
 
-# Model Card for smolvla
+# Model Card for act
 
 <!-- Provide a quick summary of what the model is/does. -->
 
 
-[SmolVLA](https://huggingface.co/papers/2506.01844) is a compact, efficient vision-language-action model that achieves competitive performance at reduced computational costs and can be deployed on consumer-grade hardware.
+[Action Chunking with Transformers (ACT)](https://huggingface.co/papers/2304.13705) is an imitation-learning method that predicts short action chunks instead of single steps. It learns from teleoperated data and often achieves high success rates.
 
 
 This policy has been trained and pushed to the Hub using [LeRobot](https://github.com/huggingface/lerobot).
@@ -32,7 +31,7 @@ Below is the short version on how to train and run inference/eval:
 ### Train from scratch
 
 ```bash
-python -m lerobot.scripts.train \
+lerobot-train \
   --dataset.repo_id=${HF_USER}/<dataset> \
   --policy.type=act \
   --output_dir=outputs/train/<desired_policy_repo_id> \
@@ -42,12 +41,12 @@ python -m lerobot.scripts.train \
   --wandb.enable=true
 ```
 
-*Writes checkpoints to `outputs/train/<desired_policy_repo_id>/checkpoints/`.*
+_Writes checkpoints to `outputs/train/<desired_policy_repo_id>/checkpoints/`._
 
 ### Evaluate the policy/run inference
 
 ```bash
-python -m lerobot.record \
+lerobot-record \
   --robot.type=so100_follower \
   --dataset.repo_id=<hf_user>/eval_<dataset> \
   --policy.path=<hf_user>/<desired_policy_repo_id> \
@@ -60,4 +59,4 @@ Prefix the dataset repo with **eval\_** and supply `--policy.path` pointing to a
 
 ## Model Details
 
-* **License:** apache-2.0
+- **License:** apache-2.0

config.json

@@ -1,12 +1,13 @@
 {
-    "type": "smolvla",
+    "type": "act",
     "n_obs_steps": 1,
-    "normalization_mapping": {
-        "VISUAL": "IDENTITY",
-        "STATE": "MEAN_STD",
-        "ACTION": "MEAN_STD"
-    },
     "input_features": {
+        "observation.state": {
+            "type": "STATE",
+            "shape": [
+                2
+            ]
+        },
         "observation.images.top": {
             "type": "VISUAL",
             "shape": [
@@ -25,50 +26,37 @@
         }
     },
     "device": "cuda",
-    "use_amp": true,
+    "use_amp": false,
     "push_to_hub": true,
     "repo_id": "smolvla",
     "private": null,
     "tags": null,
     "license": null,
-    "chunk_size": 50,
-    "n_action_steps": 50,
-    "max_state_dim": 32,
-    "max_action_dim": 32,
-    "resize_imgs_with_padding": [
-        512,
-        512
-    ],
-    "empty_cameras": 0,
-    "adapt_to_pi_aloha": false,
-    "use_delta_joint_actions_aloha": false,
-    "tokenizer_max_length": 48,
-    "num_steps": 10,
-    "use_cache": true,
-    "freeze_vision_encoder": true,
-    "train_expert_only": true,
-    "train_state_proj": true,
-    "optimizer_lr": 0.0001,
-    "optimizer_betas": [
-        0.9,
-        0.95
-    ],
-    "optimizer_eps": 1e-08,
-    "optimizer_weight_decay": 1e-10,
-    "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": false,
-    "add_image_special_tokens": false,
-    "attention_mode": "cross_attn",
-    "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,
-    "min_period": 0.004,
-    "max_period": 4.0
+    "pretrained_path": null,
+    "chunk_size": 100,
+    "n_action_steps": 100,
+    "normalization_mapping": {
+        "VISUAL": "MEAN_STD",
+        "STATE": "MEAN_STD",
+        "ACTION": "MEAN_STD"
+    },
+    "vision_backbone": "resnet18",
+    "pretrained_backbone_weights": "ResNet18_Weights.IMAGENET1K_V1",
+    "replace_final_stride_with_dilation": false,
+    "pre_norm": false,
+    "dim_model": 512,
+    "n_heads": 8,
+    "dim_feedforward": 3200,
+    "feedforward_activation": "relu",
+    "n_encoder_layers": 4,
+    "n_decoder_layers": 1,
+    "use_vae": true,
+    "latent_dim": 32,
+    "n_vae_encoder_layers": 4,
+    "temporal_ensemble_coeff": null,
+    "dropout": 0.1,
+    "kl_weight": 10.0,
+    "optimizer_lr": 1e-05,
+    "optimizer_weight_decay": 0.0001,
+    "optimizer_lr_backbone": 1e-05
 }

model.safetensors

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:171863e6c2dc85af0751d5c3a75d19d2e177b66ddc2b39bee09d8dfe4532a2db
-size 1197789712
+oid sha256:145f8bcf9e8e2ba95f132654d3940e9024907917119ecf73cceafd2b93fa7eab
+size 206675136

train_config.json

@@ -1,12 +1,12 @@
 {
     "dataset": {
-        "repo_id": "aloha/push_T_top_to_bottom_mixture_5_100percent",
+        "repo_id": "aloha/simulation_push_T_bottom_right",
         "root": null,
         "episodes": null,
         "image_transforms": {
-            "enable": true,
+            "enable": false,
             "max_num_transforms": 3,
-            "random_order": true,
+            "random_order": false,
             "tfs": {
                 "brightness": {
                     "weight": 1.0,
@@ -57,23 +57,39 @@
                             1.5
                         ]
                     }
+                },
+                "affine": {
+                    "weight": 1.0,
+                    "type": "RandomAffine",
+                    "kwargs": {
+                        "degrees": [
+                            -5.0,
+                            5.0
+                        ],
+                        "translate": [
+                            0.05,
+                            0.05
+                        ]
+                    }
                 }
             }
         },
         "revision": null,
         "use_imagenet_stats": true,
-        "video_backend": "pyav"
+        "video_backend": "pyav",
+        "streaming": false
     },
     "env": null,
     "policy": {
-        "type": "smolvla",
+        "type": "act",
         "n_obs_steps": 1,
-        "normalization_mapping": {
-            "VISUAL": "IDENTITY",
-            "STATE": "MEAN_STD",
-            "ACTION": "MEAN_STD"
-        },
         "input_features": {
+            "observation.state": {
+                "type": "STATE",
+                "shape": [
+                    2
+                ]
+            },
             "observation.images.top": {
                 "type": "VISUAL",
                 "shape": [
@@ -92,83 +108,64 @@
             }
         },
         "device": "cuda",
-        "use_amp": true,
+        "use_amp": false,
         "push_to_hub": true,
         "repo_id": "smolvla",
         "private": null,
         "tags": null,
         "license": null,
-        "chunk_size": 50,
-        "n_action_steps": 50,
-        "max_state_dim": 32,
-        "max_action_dim": 32,
-        "resize_imgs_with_padding": [
-            512,
-            512
-        ],
-        "empty_cameras": 0,
-        "adapt_to_pi_aloha": false,
-        "use_delta_joint_actions_aloha": false,
-        "tokenizer_max_length": 48,
-        "num_steps": 10,
-        "use_cache": true,
-        "freeze_vision_encoder": true,
-        "train_expert_only": true,
-        "train_state_proj": true,
-        "optimizer_lr": 0.0001,
-        "optimizer_betas": [
-            0.9,
-            0.95
-        ],
-        "optimizer_eps": 1e-08,
-        "optimizer_weight_decay": 1e-10,
-        "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": false,
-        "add_image_special_tokens": false,
-        "attention_mode": "cross_attn",
-        "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,
-        "min_period": 0.004,
-        "max_period": 4.0
+        "pretrained_path": null,
+        "chunk_size": 100,
+        "n_action_steps": 100,
+        "normalization_mapping": {
+            "VISUAL": "MEAN_STD",
+            "STATE": "MEAN_STD",
+            "ACTION": "MEAN_STD"
+        },
+        "vision_backbone": "resnet18",
+        "pretrained_backbone_weights": "ResNet18_Weights.IMAGENET1K_V1",
+        "replace_final_stride_with_dilation": false,
+        "pre_norm": false,
+        "dim_model": 512,
+        "n_heads": 8,
+        "dim_feedforward": 3200,
+        "feedforward_activation": "relu",
+        "n_encoder_layers": 4,
+        "n_decoder_layers": 1,
+        "use_vae": true,
+        "latent_dim": 32,
+        "n_vae_encoder_layers": 4,
+        "temporal_ensemble_coeff": null,
+        "dropout": 0.1,
+        "kl_weight": 10.0,
+        "optimizer_lr": 1e-05,
+        "optimizer_weight_decay": 0.0001,
+        "optimizer_lr_backbone": 1e-05
     },
-    "output_dir": "mixture_models/5augmented_smolvla/100percent/policy",
-    "job_name": "smolvla",
+    "output_dir": "sim_policy/pushT_bottom_right_smolvla",
+    "job_name": "act",
     "resume": false,
     "seed": 100000,
     "num_workers": 4,
     "batch_size": 64,
-    "steps": 30000,
+    "steps": 100000,
     "eval_freq": 20000,
     "log_freq": 200,
     "save_checkpoint": true,
-    "save_freq": 30000,
+    "save_freq": 25000,
     "use_policy_training_preset": true,
     "optimizer": {
         "type": "adamw",
-        "lr": 0.0001,
-        "weight_decay": 1e-10,
-        "grad_clip_norm": 10,
+        "lr": 1e-05,
+        "weight_decay": 0.0001,
+        "grad_clip_norm": 10.0,
         "betas": [
             0.9,
-            0.95
+            0.999
         ],
         "eps": 1e-08
     },
-    "scheduler": {
-        "type": "cosine_decay_with_warmup",
-        "num_warmup_steps": 1000,
-        "num_decay_steps": 30000,
-        "peak_lr": 0.0001,
-        "decay_lr": 2.5e-06
-    },
+    "scheduler": null,
     "eval": {
         "n_episodes": 50,
         "batch_size": 50,
@@ -180,7 +177,9 @@
         "project": "lerobot",
         "entity": null,
         "notes": null,
-        "run_id": "bqimq78k",
+        "run_id": "l9y3vem8",
         "mode": null
-    }
+    },
+    "checkpoint_path": null,
+    "rename_map": {}
 }