Nemotron Super RL is using Nano Parser

[soonchangAI]

Hi,

I noticed that the NeMo/RL configuration for Nemotron 3 Super at:

https://github.com/NVIDIA-NeMo/RL/blob/super-v3/examples/configs/super/stage1_rlvr.yaml

appears to be using the Nemotron 3 Nano reasoning parser:

http_server_serving_chat_kwargs:
  enable_auto_tools: true
  tool_parser: qwen3_coder
  reasoning_parser: nano_v3
  reasoning_parser_plugin: nemo_rl/utils/nano_v3_reasoning_parser.py

From my understanding, the reasoning formats/parsers for Nemotron 3 Nano and Nemotron 3 Super are different. Therefore, I would like to confirm whether it is expected to use the nano_v3 reasoning parser when post-training (RL) Nemotron 3 Super, or if a Super-specific reasoning parser should be used instead.

Could you clarify whether this configuration is correct?

Thank you.