update example

Author: lkevinzc

docs/reasoning_examples.md

@@ -71,7 +71,7 @@ User: Using the numbers [79, 17, 60], create an equation that equals 36. You can
 We use the following command to RL-tune a `Qwen/Qwen2.5-3B` base model and observe R1-zero-like training curves within 3 hours training on 8 GPUs.
 
 ```
-python examples/r1_zero_math.py \
+python examples/r1_zero_countdown.py \
     --critic_type grpo \
     --gpus 8 \
     --vllm_gpu_ratio 0.7 \

examples/r1_zero_math.py examples/r1_zero_countdown.pyexamples/r1_zero_math.py renamed to examples/r1_zero_countdown.py

@@ -21,8 +21,8 @@
 
 
 class ZeroMathActor(PPOActor):
-    def __init__(self, ipc_server, vllm_args, args: PPOArgs) -> None:
-        super().__init__(ipc_server, vllm_args, args)
+    def init(self, actor_id):
+        super().init(actor_id)
         if args.oracle == "countdown":
             self.oracle = CountdownOracle()
         else: