| from datasets import load_dataset, DatasetDict | |
| # Load your dataset from a local JSONL file | |
| dataset = load_dataset("json", data_files="listings.jsonl") | |
| dataset = dataset.shuffle(seed=42) | |
| # Split the dataset into train and temp | |
| split_dataset = dataset["train"].train_test_split(test_size=0.3) | |
| train_dataset = split_dataset["train"] | |
| temp_dataset = split_dataset["test"] | |
| # Split the temp dataset into validation and test | |
| split_temp_dataset = temp_dataset.train_test_split(test_size=0.5) | |
| validation_dataset = split_temp_dataset["train"] | |
| test_dataset = split_temp_dataset["test"] | |
| # Combine all splits into a DatasetDict | |
| final_dataset = DatasetDict( | |
| {"train": train_dataset, "validation": validation_dataset, "test": test_dataset} | |
| ) | |
| # Push the split datasets to Hugging Face Hub | |
| final_dataset.push_to_hub("fb-housing-posts") | |