Create dataset.py
Browse files- dataset.py +50 -0
dataset.py
ADDED
@@ -0,0 +1,50 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
import csv
|
2 |
+
from torch.utils.data import IterableDataset
|
3 |
+
from huggingface_hub import hf_hub_download
|
4 |
+
|
5 |
+
class ParsedMultiCharDataset(IterableDataset):
|
6 |
+
def __init__(self,
|
7 |
+
repo_id: str,
|
8 |
+
delimiter: str = ".,|,.",
|
9 |
+
start_file: int = 0,
|
10 |
+
num_files: int = 190,
|
11 |
+
guess_total_count = True):
|
12 |
+
self.repo_id = repo_id
|
13 |
+
self.files = [f"captions/caption_{i+start_file:03d}.csv" for i in range(num_files)]
|
14 |
+
self.delimiter = ".,|,."
|
15 |
+
self.total_rows = -1
|
16 |
+
if guess_total_count:
|
17 |
+
self.total_rows = self.guess_total_rows()
|
18 |
+
print(f"Total rows: {self.total_rows} totaling {len(self.files) * self.total_rows}")
|
19 |
+
|
20 |
+
|
21 |
+
def guess_total_rows(self):
|
22 |
+
# count the rows in the first file
|
23 |
+
path = hf_hub_download(self.repo_id, self.files[0], repo_type="dataset")
|
24 |
+
with open(path, encoding="utf-8") as f:
|
25 |
+
reader = csv.DictReader(f)
|
26 |
+
return sum(1 for _ in reader)
|
27 |
+
|
28 |
+
def __iter__(self):
|
29 |
+
for rel_path in self.files:
|
30 |
+
path = hf_hub_download(self.repo_id, rel_path, repo_type="dataset")
|
31 |
+
with open(path, encoding="utf-8") as f:
|
32 |
+
reader = csv.DictReader(f)
|
33 |
+
for row in reader:
|
34 |
+
id_ = row.get("id", "").strip()
|
35 |
+
text_field = row.get("text", "")
|
36 |
+
for caption in text_field.split(self.delimiter):
|
37 |
+
caption = caption.strip()
|
38 |
+
if caption:
|
39 |
+
yield (id_, caption)
|
40 |
+
|
41 |
+
#ds = ParsedMultiCharDataset(
|
42 |
+
# repo_id="AbstractPhil/human-templated-captions-1b",
|
43 |
+
# start_file=0,
|
44 |
+
# num_files=10
|
45 |
+
#)
|
46 |
+
#for i, ex in enumerate(ds):
|
47 |
+
# print(ex)
|
48 |
+
# if i > 10:
|
49 |
+
# break
|
50 |
+
#
|