Upload policy weights, train config and readme

README.md

@@ -1,90 +1,62 @@
 ---
-datasets:
-- lerobot/pusht
-language: en
+datasets: lerobot/pusht
 library_name: lerobot
 license: apache-2.0
+model_name: diffusion
+pipeline_tag: robotics
 tags:
-- robotics
-- imitation-learning
 - diffusion
-- mujoco
-- pytorch_model_hub_mixin
+- lerobot
+- robotics
 ---
 
-# DIFFUSION Policy — diffusion_pusht_seed3
-
-Trained with [LeRobot](https://github.com/huggingface/lerobot).  
-Date: `2026-05-28 17:14`  
-Policy type: `diffusion` | Device: `cuda`
-
----
+# Model Card for diffusion
 
-## 📦 Dataset
+<!-- Provide a quick summary of what the model is/does. -->
 
-| Parameter | Value |
-|---|---|
-| `dataset.repo_id` | `lerobot/pusht` |
 
----
+[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.
 
-## 🏋️ Training Config
 
-| Parameter | Value |
-|---|---|
-| `steps` | `7000` |
-| `batch_size` | `8` |
-| `eval_freq` | `0` |
-| `save_freq` | `2000` |
-| `num_workers` | `4` |
-| `seed` | `3` |
-| `eval.n_episodes` | `1` |
-| `eval.batch_size` | `1` |
-| `eval.use_async_envs` | `True` |
+This policy has been trained and pushed to the Hub using [LeRobot](https://github.com/huggingface/lerobot).
+See the full documentation at [LeRobot Docs](https://huggingface.co/docs/lerobot/index).
 
 ---
 
-## 📐 Policy Architecture
+## How to Get Started with the Model
 
-| Parameter | Value |
-|---|---|
-| `noise_scheduler_type` | `DDIM` |
-| `num_inference_steps` | `15` |
+For a complete walkthrough, see the [training guide](https://huggingface.co/docs/lerobot/il_robots#train-a-policy).
+Below is the short version on how to train and run inference/eval:
 
----
+### Train from scratch
 
-## 🎯 Eval Config
+```bash
+lerobot-train \
+  --dataset.repo_id=${HF_USER}/<dataset> \
+  --policy.type=act \
+  --output_dir=outputs/train/<desired_policy_repo_id> \
+  --job_name=lerobot_training \
+  --policy.device=cuda \
+  --policy.repo_id=${HF_USER}/<desired_policy_repo_id>
+  --wandb.enable=true
+```
 
-| Parameter | Value |
-|---|---|
-| `env.type` | `pusht` |
-| `env.task` | `PushT-v0` |
-| `eval.n_episodes` | `8` |
-| `eval.batch_size` | `4` |
-| `eval.use_async_envs` | `False` |
-| `policy.path` | `/kaggle/working/outputs/train/pusht_seed3/checkpoints/last/pretrained_model` |
+_Writes checkpoints to `outputs/train/<desired_policy_repo_id>/checkpoints/`._
 
----
+### Evaluate the policy/run inference
 
-## 📊 Eval Results
+```bash
+lerobot-record \
+  --robot.type=so100_follower \
+  --dataset.repo_id=<hf_user>/eval_<dataset> \
+  --policy.path=<hf_user>/<desired_policy_repo_id> \
+  --episodes=10
+```
 
-| Metric | Value |
-|---|---|
-| Episodes | `8` |
-| Success rate | `0.0%` |
-| Avg sum reward | `18.81` |
-| Avg max reward | `0.35` |
-| Eval time (s) | `52.1` |
+Prefix the dataset repo with **eval\_** and supply `--policy.path` pointing to a local or hub checkpoint.
 
 ---
 
-## Citation
+## Model Details
 
-```bibtex
-@misc{cadene2024lerobot,
-  author = {Cadene, Remi and Alibert, Simon and others},
-  title  = {LeRobot},
-  year   = {2024},
-  url    = {https://github.com/huggingface/lerobot}
-}
-```
+- **License:** apache-2.0

model.safetensors

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:7cc4844b37b6729b16ad50530ccc1f2ed634231b5ba5151488604f1a7f624a01
+oid sha256:af966f256674081411fdb62b1c626c8ad1de6225f26caebd3b7faf80a5bf0044
 size 1050861448

train_config.json

@@ -215,12 +215,12 @@
     "cudnn_deterministic": false,
     "num_workers": 4,
     "batch_size": 8,
-    "steps": 7000,
+    "steps": 70000,
     "eval_freq": 0,
     "log_freq": 200,
     "tolerance_s": 0.0001,
     "save_checkpoint": true,
-    "save_freq": 2000,
+    "save_freq": 20000,
     "use_policy_training_preset": true,
     "optimizer": {
         "type": "adam",