From d7426c7750f2839873c6512a61ec4bb27ef790f5 Mon Sep 17 00:00:00 2001
From: Shunta Saito <shunta.saito@gmail.com>
Date: Sun, 23 Feb 2025 14:47:49 +0900
Subject: [PATCH] Do not pass mask to prepare_inputs_for_generation

---
 llms/mlx_lm/models/plamo2.py | 1 -
 1 file changed, 1 deletion(-)

diff --git a/llms/mlx_lm/models/plamo2.py b/llms/mlx_lm/models/plamo2.py
index 055ccb80..da28aa2a 100644
--- a/llms/mlx_lm/models/plamo2.py
+++ b/llms/mlx_lm/models/plamo2.py
@@ -1624,7 +1624,6 @@ class Model(PlamoPreTrainedModel):
     def __call__(self, inputs: mx.array, cache: PlamoCache | None = None) -> mx.array:
         model_inputs = self.prepare_inputs_for_generation(
             input_ids=inputs,
-            attention_mask=mask,
             past_key_values=cache,
             use_cache=self.config.use_cache,
         )