From f2e1e177414bad2a50fb99b73b06d575f722a690 Mon Sep 17 00:00:00 2001 From: Paul Gauthier Date: Mon, 30 Sep 2024 14:22:11 -0700 Subject: [PATCH] avoid stomping extra_params[max_tokens]=1 in cache warming, which causes loop of 1 token infinite-output responses with prefill models #1842 #1841 --- aider/coders/base_coder.py | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/aider/coders/base_coder.py b/aider/coders/base_coder.py index 6158e7a77..2b90dbb5a 100755 --- a/aider/coders/base_coder.py +++ b/aider/coders/base_coder.py @@ -1073,7 +1073,7 @@ class Coder: self.warming_pings_left -= 1 self.next_cache_warm = time.time() + delay - kwargs = self.main_model.extra_params or dict() + kwargs = dict(self.main_model.extra_params) or dict() kwargs["max_tokens"] = 1 try: