Upload create_dataset.py with huggingface_hub

create_dataset.py

@@ -0,0 +1,94 @@
+"""
+Convert MathArena HMMT datasets to RLVR format.
+
+Usage:
+
+python scripts/data/rlvr/matharena_hmmt.py --push_to_hub
+python scripts/data/rlvr/matharena_hmmt.py --push_to_hub --hf_entity ai2-adapt-dev
+python scripts/data/rlvr/matharena_hmmt.py --dataset_field_mode math
+"""
+
+from dataclasses import dataclass
+
+import datasets
+from huggingface_hub import HfApi
+from transformers import HfArgumentParser
+
+from open_instruct import utils as open_instruct_utils
+
+SOURCE_DATASETS = (
+    ("MathArena/hmmt_feb_2025", "math_hmmt_feb_2025"),
+    ("MathArena/hmmt_nov_2025", "math_hmmt_nov_2025"),
+    ("MathArena/brumo_2025", "math_brumo_2025"),
+)
+SOURCE_DATASET_MAP = {name: label for name, label in SOURCE_DATASETS}
+
+
+@dataclass
+class Args:
+    push_to_hub: bool = False
+    hf_entity: str | None = None
+    dataset_field_mode: str = "source"
+    repo_name: str = "rlvr_matharena_2025"
+    source_dataset: str = "all"
+
+
+def _convert_dataset(source_dataset_name: str, dataset_label: str, dataset_field_mode: str) -> datasets.Dataset:
+    dataset = datasets.load_dataset(
+        source_dataset_name, split="train", num_proc=open_instruct_utils.max_num_processes()
+    )
+
+    def process(example):
+        example["messages"] = [{"role": "user", "content": example["problem"]}]
+        example["ground_truth"] = example["answer"]
+        example["dataset"] = "math" if dataset_field_mode == "math" else dataset_label
+        example["source_dataset"] = source_dataset_name
+        return example
+
+    dataset = dataset.map(process, desc=f"Converting {source_dataset_name}")
+    ordered_columns = [
+        "messages",
+        "ground_truth",
+        "dataset",
+        "source_dataset",
+        "problem_idx",
+        "problem",
+        "answer",
+        "problem_type",
+    ]
+    return dataset.select_columns([column for column in ordered_columns if column in dataset.column_names])
+
+
+def main(args: Args):
+    if args.dataset_field_mode not in {"source", "math"}:
+        raise ValueError("--dataset_field_mode must be one of: source, math")
+
+    if args.source_dataset == "all":
+        selected_datasets = SOURCE_DATASETS
+    else:
+        if args.source_dataset not in SOURCE_DATASET_MAP:
+            valid = ", ".join(sorted(SOURCE_DATASET_MAP))
+            raise ValueError(f"--source_dataset must be 'all' or one of: {valid}")
+        selected_datasets = [(args.source_dataset, SOURCE_DATASET_MAP[args.source_dataset])]
+
+    converted = [
+        _convert_dataset(source_name, dataset_label, args.dataset_field_mode) for source_name, dataset_label in selected_datasets
+    ]
+    output = datasets.DatasetDict({"train": datasets.concatenate_datasets(converted)})
+    print(output)
+
+    if args.push_to_hub:
+        api = HfApi()
+        if not args.hf_entity:
+            args.hf_entity = api.whoami()["name"]
+        repo_id = f"{args.hf_entity}/{args.repo_name}"
+        print(f"Pushing dataset to Hub: {repo_id}")
+        output.push_to_hub(repo_id)
+        api.upload_file(
+            path_or_fileobj=__file__, path_in_repo="create_dataset.py", repo_type="dataset", repo_id=repo_id
+        )
+
+
+if __name__ == "__main__":
+    parser = HfArgumentParser(Args)
+    main(*parser.parse_args_into_dataclasses())