8000 . · commit-0/build_dataset@f9f8c29 · GitHub
[go: up one dir, main page]
Skip to content

Commit f9f8c29

Browse files
wenting-zhaowenting-zhao
committed
.
1 parent 10e841d commit f9f8c29

File tree

2 files changed

+84
-0
lines changed

2 files changed

+84
-0
lines changed

‎build_simple_datasets.py

Lines changed: 72 additions & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -0,0 +1,72 @@
1+
import sys
2+
from datasets import load_dataset
3+
4+
5+
def clean_columns(dataset):
6+
keys_to_remove = [key for key in dataset.column_names if key not in ["canonical_solution", "test", "instance_id", "prompt"]]
7+
return dataset.remove_columns(keys_to_remove)
8+
9+
def convert_mbpp_tests(assert_list):
10+
# Generate individual test functions
11+
test_functions = []
12+
for i, assert_line in enumerate(assert_list, 1):
13+
test_func = f"def test{i}():\n {assert_line}"
14+
test_functions.append(test_func)
15+
16+
return "\n\n".join(test_functions)
17+
18+
def convert_humaneval_tests(test_code, entrypoint):
19+
# Split the input into lines and clean up
20+
lines = test_code.strip().split("\n")
21+
22+
# Find all assert lines
23+
assert_lines = [line for line in lines if line.lstrip().startswith("assert")]
24+
25+
# Generate individual test functions
26+
test_functions = [f"candidate = {entrypoint}"]
27+
for i, assert_line in enumerate(assert_lines, 1):
28+
test_func = f"def test{i}():\n{assert_line}"
29+
test_functions.append(test_func)
30+
31+
return "\n\n".join(test_functions)
32+
33+
def convert_humaneval():
34+
ds = load_dataset("openai/openai_humaneval")
35+
for split in ds:
36+
ds[split] = ds[split].rename_column('task_id', 'instance_id')
37+
tests = [convert_humaneval_tests(one['test'], one['entry_point']) for one in ds[split]]
38+
ds[split] = ds[split].remove_columns(['test'])
39+
ds[split] = ds[split].add_column(name='test', column=tests)
40+
ds[split] = clean_columns(ds[split])
41+
out_name = f"commit0/openai_humaneval"
42+
ds.push_to_hub(out_name)
43+
44+
def convert_codecontests():
45+
pass
46+
47+
def convert_bigcodebench():
48+
pass
49+
50+
def convert_mbpp():
51+
ds = load_dataset("google-research-datasets/mbpp")
52+
for split in ds:
53+
ds[split] = ds[split].rename_column('task_id', 'instance_id')
54+
ds[split] = ds[split].rename_column('code', 'canonical_solution')
55+
ds[split] = ds[split].add_column(name='test', column=[convert_mbpp_tests(one) for one in ds[split]['test_list']])
56+
ds[split] = clean_columns(ds[split])
57+
out_name = f"commit0/mbpp"
58+
ds.push_to_hub(out_name)
59+
60+
if __name__ == "__main__":
61+
data = sys.argv[1].lower()
62+
if data == "mbpp":
63+
convert_mbpp()
64+
elif data == "humaneval":
65+
convert_humaneval()
66+
elif data == "codecontests":
67+
convert_codecontests()
68+
elif data == "bigcodebench":
69+
convert_bigcodebench()
70+
else:
71+
raise NotImplementedError()
72+

‎get_tests.py

Lines changed: 12 additions & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -0,0 +1,12 @@
1+
import bz2
2+
from datasets import load_dataset
3+
from tqdm import tqdm
4+
5+
ds = load_dataset("princeton-nlp/SWE-bench_Verified", split="test")
6+
for one in tqdm(ds):
7+
instance_id = one["instance_id"]
8+
with bz2.open(f"tests/{instance_id}#pass_to_pass.bz2", "wt") as bz_file:
9+
bz_file.write('\n'.join(eval(one["PASS_TO_PASS"])))
10+
with bz2.open(f"tests/{instance_id}#fail_to_pass.bz2", "wt") as bz_file:
11+
bz_file.write('\n'.join(eval(one["FAIL_TO_PASS"])))
12+
print('\n'.join(one["FAIL_TO_PASS"]))

0 commit comments

Comments
 (0)
0