From 11d66c3823dd63338b239cef484ae2d2810c68d1 Mon Sep 17 00:00:00 2001
From: Frank <frank@dankware>
Date: Sun, 9 Aug 2026 22:19:32 -0700
Subject: [PATCH] Set max-model-len to 32768 for testing

---
 scripts/vllm-awq-qwen36-dflash-ampere.sh | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/scripts/vllm-awq-qwen36-dflash-ampere.sh b/scripts/vllm-awq-qwen36-dflash-ampere.sh
index 79d87c3..9835a7a 100644
--- a/scripts/vllm-awq-qwen36-dflash-ampere.sh
+++ b/scripts/vllm-awq-qwen36-dflash-ampere.sh
@@ -25,7 +25,7 @@ exec docker run --rm --name vllm-awq-qwen36-dflash-ampere \
     vllm/vllm-openai:latest "${MODEL}" \
     --served-model-name "frank-default-small" \
     --max-num-seqs 1 \
-    --max-model-len 212992 \
+    --max-model-len 32768 \
     --tensor-parallel-size 1 \
     --enable-chunked-prefill \
     --max-num-batched-tokens 4096 \