Update utils/utils.py
Browse files- utils/utils.py +6 -0
utils/utils.py
CHANGED
|
@@ -4,12 +4,16 @@ from docx import Document
|
|
| 4 |
from pptx import Presentation
|
| 5 |
from datasets import load_dataset
|
| 6 |
|
|
|
|
|
|
|
| 7 |
DATASET_TYPE_GAIA = "gaia"
|
| 8 |
DATASET_TYPE_HLE = "hle"
|
| 9 |
|
| 10 |
DATASET_FILE_PATH_GAIA = "files/gaia_validation.jsonl"
|
| 11 |
DATASET_FILE_PATH_HLE = "files/hle_validation.jsonl"
|
| 12 |
|
|
|
|
|
|
|
| 13 |
def get_dataset_from_file(dataset_type, level):
|
| 14 |
file_path = ""
|
| 15 |
|
|
@@ -58,6 +62,8 @@ def get_dataset(dataset_type, level):
|
|
| 58 |
|
| 59 |
result=[]
|
| 60 |
|
|
|
|
|
|
|
| 61 |
def is_ext(file_path, ext):
|
| 62 |
return os.path.splitext(file_path)[1].lower() == ext.lower()
|
| 63 |
|
|
|
|
| 4 |
from pptx import Presentation
|
| 5 |
from datasets import load_dataset
|
| 6 |
|
| 7 |
+
# Dataset configuration
|
| 8 |
+
|
| 9 |
DATASET_TYPE_GAIA = "gaia"
|
| 10 |
DATASET_TYPE_HLE = "hle"
|
| 11 |
|
| 12 |
DATASET_FILE_PATH_GAIA = "files/gaia_validation.jsonl"
|
| 13 |
DATASET_FILE_PATH_HLE = "files/hle_validation.jsonl"
|
| 14 |
|
| 15 |
+
# Dataset processing
|
| 16 |
+
|
| 17 |
def get_dataset_from_file(dataset_type, level):
|
| 18 |
file_path = ""
|
| 19 |
|
|
|
|
| 62 |
|
| 63 |
result=[]
|
| 64 |
|
| 65 |
+
# File processing
|
| 66 |
+
|
| 67 |
def is_ext(file_path, ext):
|
| 68 |
return os.path.splitext(file_path)[1].lower() == ext.lower()
|
| 69 |
|