From 8d9ffb3a6c1a2d47b503b1c4af98c6fcf4dadb7d Mon Sep 17 00:00:00 2001
From: Colter Downing <colter@vast.ai>
Date: Tue, 28 Oct 2025 15:28:30 -0700
Subject: [PATCH] removed 5 sec sleep and warmup request on load

---
 lib/backend.py | 10 +++++-----
 1 file changed, 5 insertions(+), 5 deletions(-)

diff --git a/lib/backend.py b/lib/backend.py
index a719b63..3ac6e33 100644
--- a/lib/backend.py
+++ b/lib/backend.py
@@ -325,10 +325,10 @@ class Backend:
                 with open(BENCHMARK_INDICATOR_FILE, "r") as f:
                     log.debug("already ran benchmark")
                     # trigger model load
-                    payload = self.benchmark_handler.make_benchmark_payload()
-                    _ = await self.__call_api(
-                        handler=self.benchmark_handler, payload=payload
-                    )
+                    # payload = self.benchmark_handler.make_benchmark_payload()
+                    # _ = await self.__call_api(
+                    #     handler=self.benchmark_handler, payload=payload
+                    # )
                     return float(f.readline())
             except FileNotFoundError:
                 pass
@@ -404,7 +404,7 @@ class Backend:
                         log.debug(f"ModelLoaded observed after {elapsed:.2f}s: {log_line}")
                         # some backends need a few seconds after logging successful startup before
                         # they can begin accepting requests
-                        await sleep(5)
+                        # await sleep(5)
                         try:
                             t_bench0 = time.time()
                             max_throughput = await run_benchmark()