fixing the eagle serving

richardhuo-nv · richardhuo-nv · commit d0198a5718a4 · 2025-08-05T17:32:26.000-07:00
diff --git a/components/backends/trtllm/engine_configs/llama4/eagle_one_model/eagle_prefill.yaml b/components/backends/trtllm/engine_configs/llama4/eagle_one_model/eagle_prefill.yaml
@@ -32,7 +32,11 @@ speculative_config:
 
 kv_cache_config:
   free_gpu_memory_fraction: 0.5
+<<<<<<< HEAD
   enable_block_reuse: false
+=======
+  enable_block_reuse: true
+>>>>>>> f8759a78b (fixing the eagle serving)
 
 cache_transceiver_config:
   backend: default
