diff --git a/convert_hf_to_gguf.py b/convert_hf_to_gguf.py
index 59410d6ce..2766fea37 100755
--- a/convert_hf_to_gguf.py
+++ b/convert_hf_to_gguf.py
@@ -2078,6 +2078,7 @@ class Phi3MiniModel(Model):
         self.gguf_writer.add_rope_dimension_count(rope_dims)
         self.gguf_writer.add_rope_freq_base(self.find_hparam(["rope_theta"]))
         self.gguf_writer.add_file_type(self.ftype)
+        self.gguf_writer.add_sliding_window(self.find_hparam(["sliding_window"]))
 
         # write rope scaling for long context (128k) model
         rope_scaling = self.find_hparam(['rope_scaling'], True)
diff --git a/src/llama.cpp b/src/llama.cpp
index 1df81eb9c..b6d6e1500 100644
--- a/src/llama.cpp
+++ b/src/llama.cpp
@@ -4974,8 +4974,7 @@ static void llm_load_hparams(
             } break;
         case LLM_ARCH_PHI3:
             {
-                hparams.n_swa = 2048;
-                ml.get_key(LLM_KV_ATTENTION_SLIDING_WINDOW, hparams.n_swa, false);
+                ml.get_key(LLM_KV_ATTENTION_SLIDING_WINDOW, hparams.n_swa);
                 ml.get_key(LLM_KV_ATTENTION_LAYERNORM_RMS_EPS, hparams.f_norm_rms_eps);
 
                 switch (hparams.n_layer) {