Upload policy weights, train config and readme

README.md

@@ -2,20 +2,20 @@
 datasets: lerobot/pusht
 library_name: lerobot
 license: apache-2.0
-model_name: act
+model_name: diffusion
 pipeline_tag: robotics
 tags:
+- diffusion
 - lerobot
 - robotics
-- act
 ---
 
-# Model Card for act
+# Model Card for diffusion
 
 <!-- Provide a quick summary of what the model is/does. -->
 
 
-[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.
+[Diffusion Policy](https://huggingface.co/papers/2303.04137) treats visuomotor control as a generative diffusion process, producing smooth, multi-step action trajectories that excel at contact-rich manipulation.
 
 
 This policy has been trained and pushed to the Hub using [LeRobot](https://github.com/huggingface/lerobot).

config.json

@@ -1,6 +1,6 @@
 {
-    "type": "act",
-    "n_obs_steps": 1,
+    "type": "diffusion",
+    "n_obs_steps": 2,
     "input_features": {
         "observation.image": {
             "type": "VISUAL",
@@ -34,30 +34,54 @@
     "tags": null,
     "license": null,
     "pretrained_path": null,
-    "chunk_size": 20,
-    "n_action_steps": 10,
+    "horizon": 32,
+    "n_action_steps": 8,
     "normalization_mapping": {
         "VISUAL": "MEAN_STD",
-        "STATE": "MEAN_STD",
-        "ACTION": "MEAN_STD"
+        "STATE": "MIN_MAX",
+        "ACTION": "MIN_MAX"
     },
+    "drop_n_last_frames": 23,
     "vision_backbone": "resnet34",
+    "resize_shape": null,
+    "crop_ratio": 1.0,
+    "crop_shape": [
+        84,
+        84
+    ],
+    "crop_is_random": true,
     "pretrained_backbone_weights": "ResNet34_Weights.IMAGENET1K_V1",
-    "replace_final_stride_with_dilation": 0,
-    "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": 2e-05,
-    "optimizer_weight_decay": 0.0001,
-    "optimizer_lr_backbone": 2e-05
+    "use_group_norm": false,
+    "spatial_softmax_num_keypoints": 32,
+    "use_separate_rgb_encoder_per_camera": false,
+    "down_dims": [
+        512,
+        1024,
+        2048
+    ],
+    "kernel_size": 5,
+    "n_groups": 8,
+    "diffusion_step_embed_dim": 128,
+    "use_film_scale_modulation": true,
+    "noise_scheduler_type": "DDIM",
+    "num_train_timesteps": 100,
+    "beta_schedule": "squaredcos_cap_v2",
+    "beta_start": 0.0001,
+    "beta_end": 0.02,
+    "prediction_type": "epsilon",
+    "clip_sample": true,
+    "clip_sample_range": 1.0,
+    "num_inference_steps": 10,
+    "compile_model": false,
+    "compile_mode": "reduce-overhead",
+    "do_mask_loss_for_padding": false,
+    "optimizer_lr": 0.0001,
+    "optimizer_betas": [
+        0.95,
+        0.999
+    ],
+    "optimizer_eps": 1e-08,
+    "optimizer_weight_decay": 1e-06,
+    "scheduler_name": "cosine",
+    "scheduler_warmup_steps": 500
 }

model.safetensors

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:88dd53e07a1bfbf8c9f089ff7de3d798168656d397ad3ac29ecf1b16a52f819e
-size 246810040
+oid sha256:17fb3860e84d4ce6c5e2a06509f143fd43fcb1c4ebc00d3f66af3bd8ee5fab29
+size 1091380504

train_config.json

@@ -81,8 +81,8 @@
     },
     "env": null,
     "policy": {
-        "type": "act",
-        "n_obs_steps": 1,
+        "type": "diffusion",
+        "n_obs_steps": 2,
         "input_features": {
             "observation.image": {
                 "type": "VISUAL",
@@ -116,35 +116,59 @@
         "tags": null,
         "license": null,
         "pretrained_path": null,
-        "chunk_size": 20,
-        "n_action_steps": 10,
+        "horizon": 32,
+        "n_action_steps": 8,
         "normalization_mapping": {
             "VISUAL": "MEAN_STD",
-            "STATE": "MEAN_STD",
-            "ACTION": "MEAN_STD"
+            "STATE": "MIN_MAX",
+            "ACTION": "MIN_MAX"
         },
+        "drop_n_last_frames": 23,
         "vision_backbone": "resnet34",
+        "resize_shape": null,
+        "crop_ratio": 1.0,
+        "crop_shape": [
+            84,
+            84
+        ],
+        "crop_is_random": true,
         "pretrained_backbone_weights": "ResNet34_Weights.IMAGENET1K_V1",
-        "replace_final_stride_with_dilation": 0,
-        "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": 2e-05,
-        "optimizer_weight_decay": 0.0001,
-        "optimizer_lr_backbone": 2e-05
+        "use_group_norm": false,
+        "spatial_softmax_num_keypoints": 32,
+        "use_separate_rgb_encoder_per_camera": false,
+        "down_dims": [
+            512,
+            1024,
+            2048
+        ],
+        "kernel_size": 5,
+        "n_groups": 8,
+        "diffusion_step_embed_dim": 128,
+        "use_film_scale_modulation": true,
+        "noise_scheduler_type": "DDIM",
+        "num_train_timesteps": 100,
+        "beta_schedule": "squaredcos_cap_v2",
+        "beta_start": 0.0001,
+        "beta_end": 0.02,
+        "prediction_type": "epsilon",
+        "clip_sample": true,
+        "clip_sample_range": 1.0,
+        "num_inference_steps": 10,
+        "compile_model": false,
+        "compile_mode": "reduce-overhead",
+        "do_mask_loss_for_padding": false,
+        "optimizer_lr": 0.0001,
+        "optimizer_betas": [
+            0.95,
+            0.999
+        ],
+        "optimizer_eps": 1e-08,
+        "optimizer_weight_decay": 1e-06,
+        "scheduler_name": "cosine",
+        "scheduler_warmup_steps": 500
     },
-    "output_dir": "/workspace/outputs/act-pusht-run11-50K",
-    "job_name": "act_pusht_run11-50K",
+    "output_dir": "/workspace/outputs/diffusion-pusht-run2-50K",
+    "job_name": "diffusion_pusht_run2-50K",
     "resume": false,
     "seed": 1000,
     "cudnn_deterministic": false,
@@ -158,17 +182,21 @@
     "save_freq": 25000,
     "use_policy_training_preset": true,
     "optimizer": {
-        "type": "adamw",
-        "lr": 2e-05,
-        "weight_decay": 0.0001,
+        "type": "adam",
+        "lr": 0.0001,
+        "weight_decay": 1e-06,
         "grad_clip_norm": 10.0,
         "betas": [
-            0.9,
+            0.95,
             0.999
         ],
         "eps": 1e-08
     },
-    "scheduler": null,
+    "scheduler": {
+        "type": "diffuser",
+        "num_warmup_steps": 500,
+        "name": "cosine"
+    },
     "eval": {
         "n_episodes": 20,
         "batch_size": 8,
@@ -180,7 +208,7 @@
         "project": "pusht-t4",
         "entity": null,
         "notes": null,
-        "run_id": "83vimi6x",
+        "run_id": "uuoy2iqs",
         "mode": null,
         "add_tags": true
     },