4ep ckpt configs + new eval screenshot + README (3-4ep optimal)

README.md

@@ -17,8 +17,8 @@ language:
 ---
 # DiffusionPolicy-PickOrange
 
-针对 [LeIsaac SO-101 PickOrange](https://github.com/LightwheelAI/leisaac) 任务**从头训练**的 LeRobot Diffusion Policy（267M，UNet 1D + ResNet18 vision encoder），**已 hot-swap 到 DDIM 32-step inference**（不重训，直接改 ckpt `config.json`）。
-_A LeRobot Diffusion Policy (267M, UNet 1D + ResNet18 vision encoder) **trained from scratch** on the [LeIsaac SO-101 PickOrange](https://github.com/LightwheelAI/leisaac) task. **DDIM 32-step inference hot-swapped into the ckpt config** without retraining._
+针对 [LeIsaac SO-101 PickOrange](https://github.com/LightwheelAI/leisaac) 任务**从头训练**的 LeRobot Diffusion Policy（267M，UNet 1D + ResNet18 vision encoder）。**本 ckpt = step-18000 ≈ 4 epoch（best）**，DDPM 100-step（如评测口径）；可选 **DDIM-32 hot-swap** 提速实时（见下，不重训、改 `config.json` 即可，且因 DDPM 太慢常撞 wall_cap，DDIM 提速后橙子率可能更高）。
+_A LeRobot Diffusion Policy (267M) **trained from scratch** on [LeIsaac SO-101 PickOrange](https://github.com/LightwheelAI/leisaac). **This ckpt = step-18000 ≈ 4 epoch (best)**, DDPM 100-step (as evaluated); optional **DDIM-32 hot-swap** for realtime (below — no retrain; DDPM is slow and often hits wall_cap, so DDIM may raise the orange rate)._
 
 ![DP eval — SO-101 PickOrange](dp-pick-orange.jpg)
 
@@ -33,11 +33,14 @@ _A LeRobot Diffusion Policy (267M, UNet 1D + ResNet18 vision encoder) **trained
   _Single-arm SO-101 picks 3 oranges sequentially and places each on a plate._
 - **数据集 / Dataset**：[`LightwheelAI/leisaac-pick-orange`](https://huggingface.co/datasets/LightwheelAI/leisaac-pick-orange) — 60 episode 遥操示范。
 - **架构 / Architecture**：Diffusion Policy（UNet 1D denoiser + ResNet18 双相机 vision encoder + 6 DOF state input → 8-step action chunk）。
-- **训练 / Training**：100k step，~1.07 GB model.safetensors，DDPM 100-step 训练。
+- **训练 / Training**：本 ckpt = **step-18000 ≈ 4 epoch**（best）。⚠️ **DP 在 60-demo 上 3–4 epoch 最优，严重过拟合即死**：早先 100k-step（≈22 ep）ckpt strict 20-round = **0/60**；退回 4 ep（本 ckpt）= **8.3% (5/60)**。~1.07 GB model.safetensors，DDPM 100-step 训练（epoch = step×8/36293）。
+  _This ckpt = **step-18000 ≈ 4 epoch** (best). DP on 60 demos peaks at **3–4 epoch**: the earlier ~22-epoch ckpt overfit to **0/60**, while 4 epoch scores **8.3% (5/60)**. Train past the sweet spot and it collapses to do-nothing._
 - **推理 hot-swap / Inference hot-swap**：`config.json` 改 `noise_scheduler_type: DDPM → DDIM` + `num_inference_steps: null → 32`，**不重训**。inference latency **393 ms → 147 ms / chunk**，slowdown 2.96x → **1.1x** 实时跑得动。
   _Edit `config.json`: `noise_scheduler_type: DDPM → DDIM` + `num_inference_steps: null → 32` — **no retraining**. Inference latency drops 393 → 147 ms/chunk, slowdown 2.96x → 1.1x, real-time on RTX 4090._
-- **评测 / Eval**：Isaac Sim 5.1 + LeIsaac，**多轮 eval 见到 0/3 ~ 3/3 全谱概率分布，部分轮能完整放完 3 颗**。Diffusion 采样自带 stochasticity，需多轮平均才有意义。
-  _Probabilistic outcomes across runs — full distribution from 0/3 to 3/3 observed, with **some rounds completing all 3 oranges**. Diffusion sampling is inherently stochastic; multi-round averaging required for meaningful comparison._
+- **评测 / Eval**（strict 20-round，leaderboard 同口径 120s/180s）：**E(🍊)/ep = 8.3% (5/60)，P(1)=25%，P(3)=0%，avg 182s**，raw `[0,1,0,1,0,0,1,0,0,0,1,0,0,0,0,0,0,0,0,1]`。本 ckpt（4 ep）能零星放第 1 颗、放不全；>15 ep 过拟合 ckpt = 0/60。
+  _Strict 20-round (leaderboard protocol): E(🍊)/ep = 8.3% (5/60), P(1)=25%, P(3)=0%, avg 182s. This 4-epoch ckpt occasionally places the 1st orange; the >15-epoch overfit ckpt scored 0/60._
+- ⚠️ **serving 坑 / serving gotcha**：DP `n_obs_steps=2` → lerobot async server 的 `predict_action_chunk` 必须 `populate_queues`（否则 stack 空 deque、server 报 `'observation.images'`、臂不动假 0）。修复在我们的 [`lerobot-v040` fork](https://github.com/vitorcen/LeIsaac-Training) 一行 patch；用官方 `pip install lerobot`（v0.5.x 未打）serve 本 ckpt 会臂不动。
+  _DP `n_obs_steps=2` needs the `populate_queues` patch in `predict_action_chunk` (else empty-deque stack → server `'observation.images'` error → frozen arm / fake 0). Patched in our lerobot fork; the stock pip lerobot will NOT serve this ckpt correctly._
 
 ## 模型亮点
 

config.json

@@ -35,13 +35,12 @@
     },
     "device": "cuda",
     "use_amp": false,
-    "use_peft": false,
     "push_to_hub": false,
     "repo_id": null,
     "private": null,
     "tags": null,
     "license": null,
-    "pretrained_path": "/home/david/work/LeIsaac/outputs/dp-leisaac-pick-orange-v2/checkpoints/060000/pretrained_model",
+    "pretrained_path": "/home/david/work/isaaclab-experience/LeIsaac/outputs/dp-grind/checkpoints/last/pretrained_model",
     "horizon": 16,
     "n_action_steps": 8,
     "normalization_mapping": {
@@ -51,11 +50,6 @@
     },
     "drop_n_last_frames": 7,
     "vision_backbone": "resnet18",
-    "resize_shape": [
-        240,
-        320
-    ],
-    "crop_ratio": 1.0,
     "crop_shape": null,
     "crop_is_random": true,
     "pretrained_backbone_weights": null,
@@ -71,7 +65,7 @@
     "n_groups": 8,
     "diffusion_step_embed_dim": 128,
     "use_film_scale_modulation": true,
-    "noise_scheduler_type": "DDIM",
+    "noise_scheduler_type": "DDPM",
     "num_train_timesteps": 100,
     "beta_schedule": "squaredcos_cap_v2",
     "beta_start": 0.0001,
@@ -79,9 +73,7 @@
     "prediction_type": "epsilon",
     "clip_sample": true,
     "clip_sample_range": 1.0,
-    "num_inference_steps": 32,
-    "compile_model": false,
-    "compile_mode": "reduce-overhead",
+    "num_inference_steps": null,
     "do_mask_loss_for_padding": false,
     "optimizer_lr": 0.0001,
     "optimizer_betas": [

train_config.json

@@ -1,7 +1,7 @@
 {
     "dataset": {
         "repo_id": "LightwheelAI/leisaac-pick-orange",
-        "root": "/home/david/work/LeIsaac/datasets/raw/leisaac-pick-orange",
+        "root": "/home/david/work/isaaclab-experience/LeIsaac/datasets/raw/leisaac-pick-orange",
         "episodes": null,
         "image_transforms": {
             "enable": false,
@@ -77,7 +77,6 @@
         "revision": null,
         "use_imagenet_stats": true,
         "video_backend": "pyav",
-        "return_uint8": false,
         "streaming": false
     },
     "env": null,
@@ -118,13 +117,12 @@
         },
         "device": "cuda",
         "use_amp": false,
-        "use_peft": false,
         "push_to_hub": false,
         "repo_id": null,
         "private": null,
         "tags": null,
         "license": null,
-        "pretrained_path": "/home/david/work/LeIsaac/outputs/dp-leisaac-pick-orange-v2/checkpoints/060000/pretrained_model",
+        "pretrained_path": "/home/david/work/isaaclab-experience/LeIsaac/outputs/dp-grind/checkpoints/last/pretrained_model",
         "horizon": 16,
         "n_action_steps": 8,
         "normalization_mapping": {
@@ -134,11 +132,6 @@
         },
         "drop_n_last_frames": 7,
         "vision_backbone": "resnet18",
-        "resize_shape": [
-            240,
-            320
-        ],
-        "crop_ratio": 1.0,
         "crop_shape": null,
         "crop_is_random": true,
         "pretrained_backbone_weights": null,
@@ -163,8 +156,6 @@
         "clip_sample": true,
         "clip_sample_range": 1.0,
         "num_inference_steps": null,
-        "compile_model": false,
-        "compile_mode": "reduce-overhead",
         "do_mask_loss_for_padding": false,
         "optimizer_lr": 0.0001,
         "optimizer_betas": [
@@ -176,21 +167,17 @@
         "scheduler_name": "cosine",
         "scheduler_warmup_steps": 500
     },
-    "output_dir": "/home/david/work/LeIsaac/outputs/dp-leisaac-pick-orange-v2",
+    "output_dir": "/home/david/work/isaaclab-experience/LeIsaac/outputs/dp-grind",
     "job_name": "diffusion",
     "resume": true,
     "seed": 1000,
-    "cudnn_deterministic": false,
-    "num_workers": 2,
-    "batch_size": 32,
-    "prefetch_factor": 4,
-    "persistent_workers": true,
-    "steps": 100000,
+    "num_workers": 0,
+    "batch_size": 8,
+    "steps": 27222,
     "eval_freq": 20000,
     "log_freq": 200,
-    "tolerance_s": 0.0001,
     "save_checkpoint": true,
-    "save_freq": 20000,
+    "save_freq": 1000,
     "use_policy_training_preset": true,
     "optimizer": {
         "type": "adam",
@@ -210,8 +197,8 @@
     },
     "eval": {
         "n_episodes": 50,
-        "batch_size": 22,
-        "use_async_envs": true
+        "batch_size": 50,
+        "use_async_envs": false
     },
     "wandb": {
         "enable": false,
@@ -220,15 +207,8 @@
         "entity": null,
         "notes": null,
         "run_id": null,
-        "mode": null,
-        "add_tags": true
+        "mode": null
     },
-    "peft": null,
-    "use_rabc": false,
-    "rabc_progress_path": null,
-    "rabc_kappa": 0.01,
-    "rabc_epsilon": 1e-06,
-    "rabc_head_mode": "sparse",
-    "rename_map": {},
-    "checkpoint_path": "/home/david/work/LeIsaac/outputs/dp-leisaac-pick-orange-v2/checkpoints/060000"
+    "checkpoint_path": "/home/david/work/isaaclab-experience/LeIsaac/outputs/dp-grind/checkpoints/last",
+    "rename_map": {}
 }