docs: add 30s dance section + AMP-vs-PPO note

README.md

@@ -19,7 +19,7 @@ pipeline_tag: reinforcement-learning
 
 # MimicKit-G1-LAFAN
 
-4 motion-tracking policies for Unitree G1 (29-DoF) trained with [MimicKit](https://github.com/xbpeng/MimicKit) DeepMimic-style PPO on 15-second LAFAN1 retargeted slices: **fight, run, dance, jumps**. Single 4090 24G, ~1 h per motion, 4096 envs × 1500 PPO iters.
+5 motion-tracking policies for Unitree G1 (29-DoF) trained with [MimicKit](https://github.com/xbpeng/MimicKit) DeepMimic-style PPO on LAFAN1 retargeted slices: **fight, run, dance, jumps** (15 s each), plus a longer **30 s dance** (`dance1_subject2`) that warm-starts from the 15 s dance and holds full-horizon tracking. Single 4090 24G, ~1 h per 15 s motion, 4096 envs.
 
 **Project repo (源码 / 复现入口):**
 - 🧪 [vitorcen/isaaclab-experience](https://github.com/vitorcen/isaaclab-experience) — MimicKit training scripts, eval chain, USD material fix, design docs
@@ -34,13 +34,18 @@ pipeline_tag: reinforcement-learning
 | **dance** | 14.70 s / **98.0 %** | 🟢 触顶 | LAFAN `dance1_subject1` [1746:2196] |
 | **jumps** | 14.70 s / **98.0 %** | 🟢 触顶 | LAFAN `jumps1_subject1` [3441:3891] |
 | **run**   | 9.45 s / **63.0 %**  | 🟡 plateau | LAFAN `run1_subject2` [3341:3791] |
+| **dance (30 s)** | full 30 s, `Test_Return` **244** (>15 s base 227) | 🟢 longer-horizon | LAFAN `dance1_subject2` [1521:2421] |
 
-3/4 motions reach ship quality at iter 1500. `run` plateaus at ~63 %, likely needing ADD-style residual or curriculum sequencing — kept as a baseline.
+3/4 of the 15 s motions reach ship quality at iter 1500. `run` plateaus at ~63 %, likely needing ADD-style residual or curriculum sequencing — kept as a baseline. The **30 s dance** doubles the horizon (900 frames @ 30 fps): warm-started from the 15 s dance ckpt and run to 2500 iters, its converged `Test_Return` (244) actually exceeds the 15 s dance (227) — the discounted return saturates near the same ceiling regardless of clip length, so matching/exceeding it means full-horizon coverage held.
 
 ---
 
 ## Demos (eval, 4 envs, with restored per-link material)
 
+**30 s dance (longer-horizon, `dance1_subject2`)** — DeepMimic PPO holds the full 30 s; an AMP baseline on the same clip could not keep rhythm (see notes below).
+
+<video controls width="640" src="https://huggingface.co/wsagi/MimicKit-G1-LAFAN/resolve/main/videos/dance1s2.mp4"></video>
+
 <video controls width="640" src="https://huggingface.co/wsagi/MimicKit-G1-LAFAN/resolve/main/videos/fight.mp4"></video>
 
 <video controls width="640" src="https://huggingface.co/wsagi/MimicKit-G1-LAFAN/resolve/main/videos/dance.mp4"></video>
@@ -72,6 +77,7 @@ Slicing rule: pick the most representative central segment of each `*1` clip (sk
 MimicKit-G1-LAFAN/
 ├── README.md
 ├── videos/
+│   ├── dance1s2.mp4 (0.7 MB, 30 s dance, first 8 s)
 │   ├── fight.mp4   (6.9 MB)
 │   ├── run.mp4     (2.6 MB)
 │   ├── dance.mp4   (1.9 MB)
@@ -83,6 +89,10 @@ MimicKit-G1-LAFAN/
 ├── run_15s/   …
 ├── dance_15s/ …
 ├── jumps_15s/ …
+├── dance_30s/                        # longer-horizon dance (dance1_subject2)
+│   ├── model.pt                      # 11 MB, final (2500 iters, warm-started)
+│   ├── env.yaml                      # 900-frame / 30 s slice
+│   └── motion.pkl
 └── assets/
     └── g1_textured.usd               # 26 MB, per-link material restored
 ```
@@ -135,7 +145,8 @@ scripts/mimickit_eval_chain.sh
 ## Known limitations
 
 - **`run` plateau at 63 %**: the LAFAN run clip has fast contact + slip; vanilla DeepMimic reward + fixed action std saturates here. Likely fixes: ADD residual, motion-curriculum from walk, or larger action std at start.
-- **15 s only**: each policy is single-clip overfit; no multi-motion conditioning. For composition, see ProtoMotions / OmniH2O.
+- **Single-clip overfit**: each policy tracks one clip; no multi-motion conditioning. For composition, see ProtoMotions / OmniH2O.
+- **AMP vs phase-tracking on dance**: an AMP (adversarial motion prior) baseline on the full 131 s `dance1_subject2` failed — the discriminator plateaued at ~0.98 agent-accuracy and the policy could not keep the choreography's rhythm. Phase-conditioned DeepMimic tracking (used here) is the right tool for high-fidelity dance; AMP fits continuous/loopable skills that don't require exact timing. The 30 s dance is the DeepMimic answer to "longer dance."
 - **No sim-to-real transfer attempted**: trained in Isaac Lab with raw observations, no domain randomization, no actuator delay model.
 
 ---