From 7c1b18ac79108d8e4e0b4222a6655579950fffb1 Mon Sep 17 00:00:00 2001
From: chenqianfzh <51831990+chenqianfzh@users.noreply.github.com>
Date: Fri, 26 Jul 2024 18:32:20 -0700
Subject: [PATCH] enforce eager mode with bnb quantization temporarily (#6846)

---
 vllm/config.py | 4 ++++
 1 file changed, 4 insertions(+)

diff --git a/vllm/config.py b/vllm/config.py
index 6403a53f86281..92fde449b43fd 100644
--- a/vllm/config.py
+++ b/vllm/config.py
@@ -282,6 +282,10 @@ def verify_with_parallel_config(
             raise ValueError(
                 "BitAndBytes quantization with TP or PP is not supported yet.")
 
+        if self.quantization == "bitsandbytes" and self.enforce_eager is False:
+            raise ValueError(
+                "BitAndBytes with enforce_eager = False is not supported yet.")
+
     def get_hf_config_sliding_window(self) -> Optional[int]:
         """Get the sliding window size, or None if disabled."""