Delete data/save_mcq.py
Browse files- data/save_mcq.py +0 -25
data/save_mcq.py
DELETED
|
@@ -1,25 +0,0 @@
|
|
| 1 |
-
from datasets import load_dataset
|
| 2 |
-
import pandas as pd
|
| 3 |
-
import os
|
| 4 |
-
|
| 5 |
-
# Load the dataset from Hugging Face
|
| 6 |
-
dataset = load_dataset("lmms-lab/MMVU", split="validation") # Change split if needed
|
| 7 |
-
|
| 8 |
-
# Convert to pandas DataFrame
|
| 9 |
-
df = dataset.to_pandas()
|
| 10 |
-
|
| 11 |
-
# Inspect columns to find question type field
|
| 12 |
-
print(df.columns)
|
| 13 |
-
print(df.head())
|
| 14 |
-
|
| 15 |
-
# Filter rows that are multiple-choice questions
|
| 16 |
-
# Adjust 'question_type' to match the actual column name in MMVU
|
| 17 |
-
mcq_df = df[df["question_type"] == "multiple-choice"]
|
| 18 |
-
|
| 19 |
-
# Save to a new .parquet file
|
| 20 |
-
output_path = "mmvu_mcq.parquet"
|
| 21 |
-
mcq_df.to_parquet(output_path, index=False)
|
| 22 |
-
|
| 23 |
-
print(f"Filtered dataset saved to {output_path}")
|
| 24 |
-
print(f"Original size: {len(df)}, MCQ size: {len(mcq_df)}")
|
| 25 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|