English

• multiturn_llm_reward.py is a file under the Slime framework for custom reward models and custom generation of rollout data, used to generate datasets and calculate reward values.

Usage:

1. Ensure that the Slime framework and related dependencies are installed.
2. Check the 【# Change Here !!!!】 comment in run_qwen2.5_3B.sh to ensure the path points to the correct file.
3. Point the reward model path in the configuration file to the multiturn_llm_reward.py file, ensuring the function name and parameters are correct.
4. Run the training script using run_qwen2.5_3B.sh to start training and evaluating the model.

To implement multi-turn + tool calling, in slime you only need to implement a custom data generation function and a reward model required for the task, corresponding to these 2 configuration items in the startup script:

CUSTOM_ARGS=(
   --custom-generate-function-path multiturn_llm_reward.generate
   --custom-rm-path multiturn_llm_reward.compute_score
)