fix: update scale down window

deployments/modal_nemotron_nano.py

@@ -64,7 +64,7 @@ app = modal.App(APP_NAME)
         "/root/.cache/huggingface": hf_cache_vol,
         "/root/.cache/vllm": vllm_cache_vol,
     },
-    scaledown_window=2 * MINUTES,
+    scaledown_window=5 * MINUTES,
     timeout=30 * MINUTES,
 )
 @modal.concurrent(max_inputs=TARGET_INPUTS)

deployments/modal_qwen36_27b_h100.py

@@ -56,7 +56,7 @@ app = modal.App(APP_NAME)
         "/root/.cache/vllm": vllm_cache_vol,
     },
     max_containers=1,
-    scaledown_window=2 * MINUTES,
+    scaledown_window=5 * MINUTES,
     timeout=30 * MINUTES,
 )
 @modal.concurrent(max_inputs=TARGET_INPUTS, target_inputs=TARGET_INPUTS)

deployments/modal_qwen36_27b_h100_llamacpp.py

@@ -48,7 +48,7 @@ app = modal.App(APP_NAME)
         "/root/.cache/llama.cpp": llama_cache_vol,
     },
     max_containers=1,
-    scaledown_window=2 * MINUTES,
+    scaledown_window=5 * MINUTES,
     timeout=30 * MINUTES,
 )
 @modal.concurrent(max_inputs=TARGET_INPUTS, target_inputs=TARGET_INPUTS)