FoundationAgents/MetaGPT · error · ValueError

Dataset {task_name} not found in config file. Available data

Error message

Dataset {task_name} not found in config file. Available datasets: {config['datasets'].keys()}

What it means

Raised by get_user_requirement when task_name is not a key in config['datasets']. It looks up the dataset's 'user_requirement' text to build the natural-language prompt for the experiment, so an unregistered task cannot produce a requirement.

Source

Thrown at metagpt/ext/sela/data/dataset.py:141

            "test_wo_target": os.path.join(data_path, "split_test_wo_target.csv"),
            "test_target": os.path.join(data_path, "split_test_target.csv"),
        }
        return split_datasets
    else:
        raise ValueError(
            f"Dataset {dataset_name} not found in config file. Available datasets: {config['datasets'].keys()}"
        )


def get_user_requirement(task_name, config):
    # datasets_dir = config["datasets_dir"]
    if task_name in config["datasets"]:
        dataset = config["datasets"][task_name]
        # data_path = os.path.join(datasets_dir, dataset["dataset"])
        user_requirement = dataset["user_requirement"]
        return user_requirement
    else:
        raise ValueError(
            f"Dataset {task_name} not found in config file. Available datasets: {config['datasets'].keys()}"
        )


def save_datasets_dict_to_yaml(datasets_dict, name="datasets.yaml"):
    with open(name, "w") as file:
        yaml.dump(datasets_dict, file)


def create_dataset_dict(dataset):
    dataset_dict = {
        "dataset": dataset.name,
        "user_requirement": dataset.create_base_requirement(),
        "metric": dataset.get_metric(),
        "target_col": dataset.target_col,
    }
    return dataset_dict

View on GitHub (pinned to 11cdf466d0)

Solutions

1. Add a 'user_requirement' (and full entry) for the task in the datasets config
2. Use a task name that appears in the 'Available datasets' list from the error
3. Verify the config object passed is the intended datasets.yaml (paths can differ per working directory)

assert task_name in config["datasets"], f"{task_name} not registered"

def is_registered_task(task_name: str, config: dict) -> bool:
    return task_name in config.get("datasets", {})

When it happens

Trigger: Calling get_user_requirement(task_name, config) with a task that has no entry in datasets.yaml, e.g. launching an experiment with --task <name> before adding that dataset.

Common situations: New/custom dataset added to the data directory but not to the yaml; task name typo; running SELA tasks from a config file of a different version.

Related errors

• Dataset {dataset_name} not found in config file. Available d
• Target column not provided
• Number of classes {num_classes} not supported
• Invalid exp_mode: {args.exp_mode}
• Dataset {task_name} not found in config file. Available data

AI-assisted analysis of FoundationAgents/MetaGPT@11cdf466d0 (2026-08-14). Data as JSON: /api/errors/35697187100570d0. Report an issue: GitHub.