Datasets:

---

datapizza-ai-lab

/

salaries

like 56

Follow

Datapizza Labs 21

Dataset schema inconsistency: salary declared as int but contains float values

Description

While loading the dataset datapizza-ai-lab/salaries using datasets.load_dataset, the dataset generation fails with an ArrowInvalid error related to type casting.

The root cause is an inconsistency between the declared schema and the actual data:

• The schema declares salary as int
• The data contains float values, including non-integer values (e.g. with decimals)

This causes Apache Arrow to fail when casting float → int64.

Error

ArrowInvalid: Float value 28546.310000 was truncated converting to int64

The error occurs during load_dataset()

Example problematic records

Row n 25370

{
  ...
  "salary":28546.31,
  ...
}

Suggested fix

One of the following (preferred option first):

Update the schema to:

salary  float   RAL (Gross Annual Salary) in EUR

Alternatively, normalize the data so that salary is always an integer (e.g. rounding), if that is the intended semantics.

Workaround (for users)

features = Features({
    "jobTitle": Value("string"),
    "monthsOfExperience": Value("int64"),
    "educationType": Value("string"),
    "companyIndustry": Value("string"),
    "companySize": Value("string"),
    "province": Value("string"),
    "workMode": Value("string"),
    "technologies": Sequence(Value("string")),
    "salary": Value("float64"),          # 👈 FIX
    "age": Value("int64"),
    "gender": Value("string"),
    "aiUsageFrequency": Value("int64"),
    "aiTechnologies": Sequence(Value("string")),
    "aiTaskTypes": Sequence(Value("string")),
    "aiUpdateSources": Sequence(Value("string")),
    "admiredTechnologies": Sequence(Value("string")),
    "editor": Value("string"),
    "rating": Value("int64"),
    "valid": Value("bool"),
    "submittedAt": Value("string"),
})

dataset = load_dataset(
    "datapizza-ai-lab/salaries",
    features=features
)