flagging to datasets
Browse files- app.py +27 -58
- bad_examples/reddit_threaded_bad_examples.jsonl +2 -2
- report.jsonl +3 -0
- requirements.txt +2 -1
app.py
CHANGED
@@ -1,61 +1,39 @@
|
|
1 |
import json
|
2 |
import math
|
|
|
|
|
3 |
from functools import partial
|
4 |
|
|
|
5 |
import streamlit as st
|
6 |
import streamlit.components.v1 as components
|
7 |
-
from
|
8 |
|
9 |
BAD_EXAMPLES_PATH = "bad_examples"
|
10 |
DATA_PATH = "data"
|
11 |
-
|
12 |
-
|
13 |
-
|
14 |
-
|
15 |
-
|
16 |
-
|
17 |
-
|
18 |
-
|
19 |
-
|
20 |
-
|
21 |
-
|
22 |
-
|
23 |
-
|
24 |
-
|
25 |
-
|
26 |
-
|
27 |
-
|
28 |
-
|
29 |
-
|
30 |
-
|
31 |
-
|
32 |
-
|
33 |
-
return metadata
|
34 |
-
if element.name == "Flagging Reason":
|
35 |
-
return reason
|
36 |
-
if element.name == "Flagging Person":
|
37 |
-
return person
|
38 |
-
if element.name == "Part":
|
39 |
-
return part
|
40 |
-
|
41 |
-
|
42 |
-
def report_result(dataset, docid, text, metadata, reason, person, part):
|
43 |
-
form = Form()
|
44 |
-
FORM_URL = "https://docs.google.com/forms/d/e/1FAIpQLSedYTj1pBD5L6xo6qPUKY5vleNW183FXCgc3LSSgg3AUwZWKA/viewform"
|
45 |
-
form.load(FORM_URL)
|
46 |
-
form.fill(
|
47 |
-
partial(
|
48 |
-
form_callback,
|
49 |
-
dataset=dataset,
|
50 |
-
docid=docid,
|
51 |
-
text=text,
|
52 |
-
metadata=metadata,
|
53 |
-
reason=reason,
|
54 |
-
person=person,
|
55 |
-
part=part,
|
56 |
-
),
|
57 |
)
|
58 |
-
form.submit()
|
59 |
|
60 |
|
61 |
def load_jsonl(file_path):
|
@@ -84,7 +62,6 @@ def save_flag_and_get_next_item(sample, issue):
|
|
84 |
f.write(json.dumps(sample) + "\n")
|
85 |
|
86 |
text = sample["text"]
|
87 |
-
|
88 |
sample.pop("text")
|
89 |
sample.pop("issue")
|
90 |
sample_id = ""
|
@@ -94,15 +71,7 @@ def save_flag_and_get_next_item(sample, issue):
|
|
94 |
else:
|
95 |
sample_id = sample["id"]
|
96 |
|
97 |
-
|
98 |
-
num_parts = math.ceil(len(text) / MAX_DOC_LENGTH)
|
99 |
-
for i in range(num_parts):
|
100 |
-
text_portion = text[i * MAX_DOC_LENGTH : (i + 1) * MAX_DOC_LENGTH]
|
101 |
-
report_result(
|
102 |
-
dataset, sample_id, text_portion, str(sample), issue, "", str(i)
|
103 |
-
)
|
104 |
-
else:
|
105 |
-
report_result(dataset, sample_id, text, str(sample), issue, "", str(0))
|
106 |
|
107 |
get_next_item()
|
108 |
|
|
|
1 |
import json
|
2 |
import math
|
3 |
+
import os
|
4 |
+
import uuid
|
5 |
from functools import partial
|
6 |
|
7 |
+
import jsonlines
|
8 |
import streamlit as st
|
9 |
import streamlit.components.v1 as components
|
10 |
+
from huggingface_hub import HfApi
|
11 |
|
12 |
BAD_EXAMPLES_PATH = "bad_examples"
|
13 |
DATA_PATH = "data"
|
14 |
+
|
15 |
+
|
16 |
+
def report_result_dataset(dataset, docid, text, metadata, reason, annotator):
|
17 |
+
with jsonlines.open("report.jsonl", "w") as f:
|
18 |
+
f.write(
|
19 |
+
{
|
20 |
+
"dataset": dataset,
|
21 |
+
"docid": docid,
|
22 |
+
"text": text,
|
23 |
+
"metadata": metadata,
|
24 |
+
"reason": reason,
|
25 |
+
"annotator": annotator,
|
26 |
+
}
|
27 |
+
)
|
28 |
+
|
29 |
+
api = HfApi()
|
30 |
+
api.upload_file(
|
31 |
+
path_or_fileobj="report.jsonl",
|
32 |
+
path_in_repo="report-{}.jsonl".format(uuid.uuid4()),
|
33 |
+
repo_id="HuggingFaceGECLM/data_feedback",
|
34 |
+
repo_type="dataset",
|
35 |
+
token=os.environ.get("geclm_token"),
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
36 |
)
|
|
|
37 |
|
38 |
|
39 |
def load_jsonl(file_path):
|
|
|
62 |
f.write(json.dumps(sample) + "\n")
|
63 |
|
64 |
text = sample["text"]
|
|
|
65 |
sample.pop("text")
|
66 |
sample.pop("issue")
|
67 |
sample_id = ""
|
|
|
71 |
else:
|
72 |
sample_id = sample["id"]
|
73 |
|
74 |
+
report_result_dataset(dataset, sample_id, text, str(sample), issue, "")
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
75 |
|
76 |
get_next_item()
|
77 |
|
bad_examples/reddit_threaded_bad_examples.jsonl
CHANGED
@@ -1,3 +1,3 @@
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
-
oid sha256:
|
3 |
-
size
|
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:562ed8ca881c564329b0cb138863cdedfd8339913635ddd1f4c733fc723b3230
|
3 |
+
size 13634
|
report.jsonl
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:c93030e14be30172f1313437e54c47c138d99b91cedefc965f88ba1e5c6025c6
|
3 |
+
size 2250
|
requirements.txt
CHANGED
@@ -1,2 +1,3 @@
|
|
1 |
-
|
|
|
2 |
streamlit==1.20.0
|
|
|
1 |
+
huggingface_hub
|
2 |
+
jsonlines
|
3 |
streamlit==1.20.0
|