Shorter GPU reservation (more free runs per visitor); first-run and quota note

app.py

@@ -92,7 +92,7 @@ def parse_table(text, lang):
     return qs
 
 
-@spaces.GPU(duration=20, size=GPU_SIZE)
+@spaces.GPU(duration=6, size=GPU_SIZE)  # a call takes ~0.3 s; a short reservation leaves visitors more free runs
 def _run(x, qs, n_perm):
     t = time.perf_counter()
     res = DECIDER.decide(x, qs) if MOCK else DECIDER.decide(x, qs, n_perm=n_perm)
@@ -145,7 +145,11 @@ finished? should I stop talking? are they angry?* Every answer comes from **one
 10 decisions in 92 ms, 21× faster than ASR → LLM ([paper](https://arxiv.org/abs/2610.02638)).
 
 ⭐ [GitHub](https://github.com/adventists-ai/duplexjev) · 🤗 [Models](https://huggingface.co/adventists-ai) ·
-🔌 Pipecat: `pip install pipecat-duplexjev` · 🏢 Commercial licence & hosted API: **jiejin@adventists.ai**
+🔌 Pipecat: `pip install pipecat-duplexjev` · LiveKit: `pip install livekit-plugins-duplexjev` · 🏢 Commercial licence & hosted API: **jiejin@adventists.ai**
+
+<sub>⏱️ The first run after the Space has been idle takes a few seconds while Hugging Face attaches a GPU; after that each run takes about 0.3 s on the GPU.
+Free GPU time per visitor is limited by Hugging Face — [log in](https://huggingface.co/login) for more runs.
+首次运行需要几秒分配 GPU，之后每次约 0.3 秒；登录 Hugging Face 可获得更多免费次数。</sub>
 """
 
 FOOT = """