[neo] Correct enforce_eager default for LMI sharding (#2618)

deepjavalibrary · Dec 2, 2024 · b0b86c2 · b0b86c2
1 parent 6a29680
commit b0b86c2
Showing 1 changed file with 1 addition and 1 deletion.
diff --git a/serving/docker/partition/sm_neo_shard.py b/serving/docker/partition/sm_neo_shard.py
@@ -108,7 +108,7 @@ def shard_lmi_dist_model(self, input_dir: str, output_dir: str,
         gpu_memory_utilization = float(
             self.properties.get("option.gpu_memory_utilization", 0.9))
         enforce_eager: bool = self.properties.get("option.enforce_eager",
-                                                  "true").lower() == "true"
+                                                  "false").lower() == "true"
         max_rolling_batch_size = int(
             self.properties.get("option.max_rolling_batch_size", 256))
         max_model_len = self.properties.get("option.max_model_len", None)