From 82dfd39ef255f79b273f40452588c72675fbe1b4 Mon Sep 17 00:00:00 2001 From: Awni Hannun Date: Mon, 8 Dec 2025 16:14:00 -0800 Subject: [PATCH] default repetition penalty to 0.0 in the server (#658) --- mlx_lm/server.py | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/mlx_lm/server.py b/mlx_lm/server.py index fa4c307..c109ad8 100644 --- a/mlx_lm/server.py +++ b/mlx_lm/server.py @@ -921,7 +921,7 @@ class APIHandler(BaseHTTPRequestHandler): self.top_p = self.body.get("top_p", self.response_generator.cli_args.top_p) self.top_k = self.body.get("top_k", self.response_generator.cli_args.top_k) self.min_p = self.body.get("min_p", self.response_generator.cli_args.min_p) - self.repetition_penalty = self.body.get("repetition_penalty", 1.0) + self.repetition_penalty = self.body.get("repetition_penalty", 0.0) self.repetition_context_size = self.body.get("repetition_context_size", 20) self.xtc_probability = self.body.get("xtc_probability", 0.0) self.xtc_threshold = self.body.get("xtc_threshold", 0.0)