Upload policy weights, train config and readme

README.md

@@ -1,23 +1,29 @@
 ---
-base_model: lerobot/smolvla_base
 datasets: GetSoloTech/FoodStack
 library_name: lerobot
 license: apache-2.0
-model_name: smolvla
+model_name: pi05
 pipeline_tag: robotics
 tags:
-- lerobot
+- pi05
 - robotics
-- smolvla
-- unsloth
+- lerobot
 ---
 
-# Model Card for smolvla
+# Model Card for pi05
 
 <!-- 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.
+**π₀.₅ (Pi05) Policy**
+
+π₀.₅ is a Vision-Language-Action model with open-world generalization, from Physical Intelligence. The LeRobot implementation is adapted from their open source OpenPI repository.
+
+**Model Overview**
+
+π₀.₅ represents a significant evolution from π₀, developed by Physical Intelligence to address a big challenge in robotics: open-world generalization. While robots can perform impressive tasks in controlled environments, π₀.₅ is designed to generalize to entirely new environments and situations that were never seen during training.
+
+For more details, see the [Physical Intelligence π₀.₅ blog post](https://www.physicalintelligence.company/blog/pi05).
 
 
 This policy has been trained and pushed to the Hub using [LeRobot](https://github.com/huggingface/lerobot).

config.json

@@ -1,5 +1,5 @@
 {
-    "type": "smolvla",
+    "type": "pi05",
     "n_obs_steps": 1,
     "input_features": {
         "observation.state": {
@@ -40,50 +40,44 @@
     "private": null,
     "tags": null,
     "license": null,
-    "pretrained_path": "lerobot/smolvla_base",
+    "pretrained_path": "lerobot/pi05_base",
+    "paligemma_variant": "gemma_2b",
+    "action_expert_variant": "gemma_300m",
+    "dtype": "bfloat16",
     "chunk_size": 50,
     "n_action_steps": 50,
-    "normalization_mapping": {
-        "VISUAL": "IDENTITY",
-        "STATE": "MEAN_STD",
-        "ACTION": "MEAN_STD"
-    },
     "max_state_dim": 32,
     "max_action_dim": 32,
-    "resize_imgs_with_padding": [
-        512,
-        512
+    "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,
-    "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,
+    "tokenizer_max_length": 200,
+    "normalization_mapping": {
+        "ACTION": "MEAN_STD",
+        "STATE": "MEAN_STD",
+        "VISUAL": "IDENTITY"
+    },
+    "gradient_checkpointing": true,
+    "compile_model": true,
+    "compile_mode": "max-autotune",
+    "optimizer_lr": 2.5e-05,
     "optimizer_betas": [
         0.9,
         0.95
     ],
     "optimizer_eps": 1e-08,
-    "optimizer_weight_decay": 1e-10,
-    "optimizer_grad_clip_norm": 10,
+    "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,
-    "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
+    "scheduler_decay_lr": 2.5e-06
 }

model.safetensors

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:abf6d4335765e784ba6ffb392d36e51cb0f97de44327ec6d1b351d5c5c4919b7
-size 1197789224
+oid sha256:3e9f0192d2f24b447d95d1ec4e23b8da5e8972faae87207c0e7e09af9e094b56
+size 7473096344

train_config.json

@@ -81,7 +81,7 @@
     },
     "env": null,
     "policy": {
-        "type": "smolvla",
+        "type": "pi05",
         "n_obs_steps": 1,
         "input_features": {
             "observation.state": {
@@ -122,55 +122,49 @@
         "private": null,
         "tags": null,
         "license": null,
-        "pretrained_path": "lerobot/smolvla_base",
+        "pretrained_path": "lerobot/pi05_base",
+        "paligemma_variant": "gemma_2b",
+        "action_expert_variant": "gemma_300m",
+        "dtype": "bfloat16",
         "chunk_size": 50,
         "n_action_steps": 50,
-        "normalization_mapping": {
-            "VISUAL": "IDENTITY",
-            "STATE": "MEAN_STD",
-            "ACTION": "MEAN_STD"
-        },
         "max_state_dim": 32,
         "max_action_dim": 32,
-        "resize_imgs_with_padding": [
-            512,
-            512
+        "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,
-        "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,
+        "tokenizer_max_length": 200,
+        "normalization_mapping": {
+            "ACTION": "MEAN_STD",
+            "STATE": "MEAN_STD",
+            "VISUAL": "IDENTITY"
+        },
+        "gradient_checkpointing": true,
+        "compile_model": true,
+        "compile_mode": "max-autotune",
+        "optimizer_lr": 2.5e-05,
         "optimizer_betas": [
             0.9,
             0.95
         ],
         "optimizer_eps": 1e-08,
-        "optimizer_weight_decay": 1e-10,
-        "optimizer_grad_clip_norm": 10,
+        "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,
-        "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
+        "scheduler_decay_lr": 2.5e-06
     },
     "output_dir": "/workspace/outputs",
-    "job_name": "uhyugtu",
+    "job_name": "qqqqqqqqqqqqqqqqqvlaaa",
     "resume": false,
     "seed": 1000,
     "num_workers": 4,
@@ -179,13 +173,13 @@
     "eval_freq": 20000,
     "log_freq": 200,
     "save_checkpoint": true,
-    "save_freq": 3000,
+    "save_freq": 1000,
     "use_policy_training_preset": true,
     "optimizer": {
         "type": "adamw",
-        "lr": 0.0001,
-        "weight_decay": 1e-10,
-        "grad_clip_norm": 10,
+        "lr": 2.5e-05,
+        "weight_decay": 0.01,
+        "grad_clip_norm": 1.0,
         "betas": [
             0.9,
             0.95
@@ -196,7 +190,7 @@
         "type": "cosine_decay_with_warmup",
         "num_warmup_steps": 1000,
         "num_decay_steps": 30000,
-        "peak_lr": 0.0001,
+        "peak_lr": 2.5e-05,
         "decay_lr": 2.5e-06
     },
     "eval": {