You can deploy this model and start earning money today!

fengyao1909/
dapo-math-17k-deduplicated

Views

No views yet

dapo dataset processed with community instructions. import pandas as pd import polars as pl df = pd.read_parquet('DAPO-Math-17k/data/dapo-math-17k.parquet')

Remove duplicates

pl_df = pl.from_pandas(df).unique(subset=["data_source", "prompt", "ability", "reward_model"])

Count number of reward_models per prompt

pl_df = pl_df.with_columns( pl.col("reward_model").n_unique().over("prompt").alias("n_rm") )

Keep only prompts with one reward_model

cleaned = pl_df.filter(pl.col("n_rm") ==… See the full description on the dataset page: https://huggingface.co/datasets/fengyao1909/dapo-math-17k-deduplicated.