I want a command-line converter for preparing GSM8K data for MathRuler.
I want a command-line converter for preparing GSM8K data for MathRuler.: a task in MiMo-V2.6-RL-oss: Agentic RL Environments (MiMo RL release). I should be able to run python scripts/gsm8k to math.py <data folder , where <data folder contains train.jsonl and test.jsonl from the original GSM8K…
The task
I should be able to run python scripts/gsm8k_to_math.py <data_folder>, where <data_folder> contains train.jsonl and test.jsonl from the original GSM8K release, and the command should create data/gsm8k_splits/ if needed and write normalized train.jsonl and test.jsonl files…
Part of XiaomiMiMo/MiMo-V2.6-RL-oss.