Upload policy weights, train config and readme

README.md

@@ -2,20 +2,20 @@
 datasets: data/banana_plate
 library_name: lerobot
 license: apache-2.0
-model_name: pi0_enhanced
+model_name: act
 pipeline_tag: robotics
 tags:
-- robotics
+- act
 - lerobot
-- pi0_enhanced
+- robotics
 ---
 
-# Model Card for pi0_enhanced
+# Model Card for act
 
 <!-- Provide a quick summary of what the model is/does. -->
 
 
-_Model type not recognized — please update this template._
+[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).

config.json

@@ -1,5 +1,5 @@
 {
-    "type": "pi0_enhanced",
+    "type": "act",
     "n_obs_steps": 1,
     "input_features": {
         "observation.state": {
@@ -49,61 +49,30 @@
     "tags": null,
     "license": null,
     "pretrained_path": null,
-    "paligemma_variant": "gemma_2b",
-    "action_expert_variant": "gemma_300m",
-    "dtype": "bfloat16",
-    "chunk_size": 50,
-    "n_action_steps": 50,
-    "max_state_dim": 32,
-    "max_action_dim": 32,
-    "num_inference_steps": 10,
-    "time_sampling_beta_alpha": 1.5,
-    "time_sampling_beta_beta": 1.0,
-    "time_sampling_scale": 0.999,
-    "time_sampling_offset": 0.001,
-    "min_period": 0.004,
-    "max_period": 4.0,
-    "image_resolution": [
-        224,
-        224
-    ],
-    "empty_cameras": 0,
-    "freeze_vision_encoder": true,
-    "msap_enabled": true,
-    "msap_scales": [
-        1,
-        2,
-        4
-    ],
-    "msap_num_heads": 8,
-    "msap_reduction_factor": 4,
-    "scroi_enabled": true,
-    "scroi_num_roi_queries": 8,
-    "scroi_num_heads": 8,
-    "had_enabled": true,
-    "had_coarse_factor": 5,
-    "had_num_transformer_layers": 2,
-    "had_num_heads": 4,
-    "had_use_auxiliary_loss": true,
-    "had_auxiliary_loss_weight": 0.1,
+    "chunk_size": 100,
+    "n_action_steps": 100,
     "normalization_mapping": {
-        "VISUAL": "IDENTITY",
+        "VISUAL": "MEAN_STD",
         "STATE": "MEAN_STD",
         "ACTION": "MEAN_STD"
     },
-    "gradient_checkpointing": true,
-    "compile_model": false,
-    "compile_mode": "max-autotune",
-    "optimizer_lr": 2.5e-05,
-    "optimizer_betas": [
-        0.9,
-        0.95
-    ],
-    "optimizer_eps": 1e-08,
-    "optimizer_weight_decay": 0.01,
-    "optimizer_grad_clip_norm": 1.0,
-    "scheduler_warmup_steps": 1000,
-    "scheduler_decay_steps": 30000,
-    "scheduler_decay_lr": 2.5e-06,
-    "tokenizer_max_length": 48
+    "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:bf1758948fc91d6dfd1da0b25d252f1942421842bd20d4a6a1bceaa88f24e7ef
-size 7220225208
+oid sha256:29cc908fd5d4d0a458a5fdd89888e0b61e035532db52691bdc599896762db3a2
+size 206748912

train_config.json

@@ -81,7 +81,7 @@
     },
     "env": null,
     "policy": {
-        "type": "pi0_enhanced",
+        "type": "act",
         "n_obs_steps": 1,
         "input_features": {
             "observation.state": {
@@ -131,71 +131,40 @@
         "tags": null,
         "license": null,
         "pretrained_path": null,
-        "paligemma_variant": "gemma_2b",
-        "action_expert_variant": "gemma_300m",
-        "dtype": "bfloat16",
-        "chunk_size": 50,
-        "n_action_steps": 50,
-        "max_state_dim": 32,
-        "max_action_dim": 32,
-        "num_inference_steps": 10,
-        "time_sampling_beta_alpha": 1.5,
-        "time_sampling_beta_beta": 1.0,
-        "time_sampling_scale": 0.999,
-        "time_sampling_offset": 0.001,
-        "min_period": 0.004,
-        "max_period": 4.0,
-        "image_resolution": [
-            224,
-            224
-        ],
-        "empty_cameras": 0,
-        "freeze_vision_encoder": true,
-        "msap_enabled": true,
-        "msap_scales": [
-            1,
-            2,
-            4
-        ],
-        "msap_num_heads": 8,
-        "msap_reduction_factor": 4,
-        "scroi_enabled": true,
-        "scroi_num_roi_queries": 8,
-        "scroi_num_heads": 8,
-        "had_enabled": true,
-        "had_coarse_factor": 5,
-        "had_num_transformer_layers": 2,
-        "had_num_heads": 4,
-        "had_use_auxiliary_loss": true,
-        "had_auxiliary_loss_weight": 0.1,
+        "chunk_size": 100,
+        "n_action_steps": 100,
         "normalization_mapping": {
-            "VISUAL": "IDENTITY",
+            "VISUAL": "MEAN_STD",
             "STATE": "MEAN_STD",
             "ACTION": "MEAN_STD"
         },
-        "gradient_checkpointing": true,
-        "compile_model": false,
-        "compile_mode": "max-autotune",
-        "optimizer_lr": 2.5e-05,
-        "optimizer_betas": [
-            0.9,
-            0.95
-        ],
-        "optimizer_eps": 1e-08,
-        "optimizer_weight_decay": 0.01,
-        "optimizer_grad_clip_norm": 1.0,
-        "scheduler_warmup_steps": 1000,
-        "scheduler_decay_steps": 30000,
-        "scheduler_decay_lr": 2.5e-06,
-        "tokenizer_max_length": 48
+        "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": "outputs/test_pi0_enhanced",
-    "job_name": "pi0_enhanced",
+    "output_dir": "/media/choi/HDD/outputs/train/ACT_enhanced_banana",
+    "job_name": "pi0_enhanced_banana",
     "resume": false,
     "seed": 1000,
-    "num_workers": 0,
-    "batch_size": 1,
-    "steps": 10,
+    "num_workers": 4,
+    "batch_size": 8,
+    "steps": 100000,
     "eval_freq": 20000,
     "log_freq": 200,
     "save_checkpoint": true,
@@ -203,22 +172,16 @@
     "use_policy_training_preset": true,
     "optimizer": {
         "type": "adamw",
-        "lr": 2.5e-05,
-        "weight_decay": 0.01,
-        "grad_clip_norm": 1.0,
+        "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": 2.5e-05,
-        "decay_lr": 2.5e-06
-    },
+    "scheduler": null,
     "eval": {
         "n_episodes": 50,
         "batch_size": 50,