diff --git a/example_trainer/scripts/compare_all_modes_math_zero.sh b/example_trainer/scripts/compare_all_modes_math_zero.sh
index 2c1f511e..55065d54 100755
--- a/example_trainer/scripts/compare_all_modes_math_zero.sh
+++ b/example_trainer/scripts/compare_all_modes_math_zero.sh
@@ -244,7 +244,7 @@ python -u environments/math_server_zero.py serve \
 SHARED_ENV_PID=$!
 
 echo "[SHARED_VLLM] Starting trainer..."
-CUDA_VISIBLE_DEVICES=$SHARED_GPU python -m example_trainer.grpo \
+CUDA_VISIBLE_DEVICES=$SHARED_GPU PYTHONUNBUFFERED=1 stdbuf -oL -eL python -u -m example_trainer.grpo \
     --model-name "$MODEL" \
     --weight-bridge-mode shared_vllm \
     --vllm-port $SHARED_VLLM_PORT \
@@ -277,7 +277,7 @@ python -u environments/math_server_zero.py serve \
 LORA_ONLY_ENV_PID=$!
 
 echo "[LORA_ONLY] Starting trainer..."
-CUDA_VISIBLE_DEVICES=$LORA_ONLY_GPU python -m example_trainer.grpo \
+CUDA_VISIBLE_DEVICES=$LORA_ONLY_GPU PYTHONUNBUFFERED=1 stdbuf -oL -eL python -u -m example_trainer.grpo \
     --model-name "$MODEL" \
     --weight-bridge-mode lora_only \
     --vllm-port $LORA_ONLY_VLLM_PORT \
@@ -301,7 +301,8 @@ LORA_ONLY_TRAINER_PID=$!
 echo ""
 echo "[LORA_RESTART] Starting trainer (manages vLLM internally)..."
 # NOTE: lora_restart shares GPU with trainer's model (~8GB), so use lower vLLM memory
-CUDA_VISIBLE_DEVICES=$LORA_RESTART_GPU python -m example_trainer.grpo \
+# Use unbuffered output (-u) and stdbuf to capture crashes
+CUDA_VISIBLE_DEVICES=$LORA_RESTART_GPU PYTHONUNBUFFERED=1 stdbuf -oL -eL python -u -m example_trainer.grpo \
     --model-name "$MODEL" \
     --weight-bridge-mode lora_restart \
     --vllm-port $LORA_RESTART_VLLM_PORT \
diff --git a/example_trainer/vllm_manager.py b/example_trainer/vllm_manager.py
index 4088994c..7f9f20fe 100644
--- a/example_trainer/vllm_manager.py
+++ b/example_trainer/vllm_manager.py
@@ -45,10 +45,10 @@ def kill_process_on_port(port: int, timeout: float = 5.0) -> bool:
         )
         if result.stdout.strip():
             pids = result.stdout.strip().split("\n")
+            print(f"  Killing {len(pids)} processes on port {port}...")
             for pid in pids:
                 try:
                     os.kill(int(pid), signal.SIGTERM)
-                    print(f"  Sent SIGTERM to PID {pid}")
                 except (ProcessLookupError, ValueError):
                     pass
 
@@ -61,12 +61,15 @@ def kill_process_on_port(port: int, timeout: float = 5.0) -> bool:
                 time.sleep(0.5)
 
             # Force kill if still running
+            killed_count = 0
             for pid in pids:
                 try:
                     os.kill(int(pid), signal.SIGKILL)
-                    print(f"  Sent SIGKILL to PID {pid}")
+                    killed_count += 1
                 except (ProcessLookupError, ValueError):
                     pass
+            if killed_count > 0:
+                print(f"  Force killed {killed_count} stubborn processes")
 
             time.sleep(1)
             return not is_port_in_use(port)