use top p in server

llms/mlx_lm/sample_utils.py

@@ -190,7 +190,7 @@ def make_repetition_penalty(penalty: float, context_size: int = 20):
         Callable[[mx.array, List[int]], mx.array]:
             The repetition penalty processor.
     """
-    if penalty < 0 or not isinstance(penalty, float):
+    if penalty < 0 or not isinstance(penalty, (int, float)):
         raise ValueError(f"penalty must be a non-negative float, got {penalty}")
 
     def repetition_penalty_processor(tokens, logits):

llms/mlx_lm/server.py

@@ -465,7 +465,7 @@ class APIHandler(BaseHTTPRequestHandler):
 
         text = ""
         tic = time.perf_counter()
-        sampler = make_sampler(self.temperature)
+        sampler = make_sampler(self.temperature, top_p=self.top_p)
         logits_processors = make_logits_processors(
             self.logit_bias, self.repetition_penalty, self.repetition_context_size
         )