SERA: Soft-Verified Efficient Repository Agents
Paper • 2601.20789 • Published • 13
path stringlengths 15 15 | task_binary unknown | bug_label stringclasses 51
values | source_dataset stringclasses 1
value |
|---|---|---|---|
swesmith-000000 | [
31,
139,
8,
0,
0,
0,
0,
0,
2,
255,
237,
125,
123,
119,
227,
54,
146,
111,
255,
173,
79,
129,
113,
246,
140,
237,
109,
73,
150,
228,
103,
60,
235,
73,
156,
182,
39,
237,
187,
73,
187,
215,
237,
206,
156,
156,
116,
14,
3,
... | efficiency_resources | DCAgent/swesmith-sandboxes-with_tests-25k |
swesmith-000001 | [
31,
139,
8,
0,
0,
0,
0,
0,
2,
255,
237,
125,
123,
119,
220,
54,
178,
167,
255,
238,
79,
129,
233,
220,
51,
146,
142,
37,
246,
67,
207,
40,
163,
185,
81,
108,
39,
241,
78,
98,
249,
218,
206,
228,
238,
137,
125,
24,
54,
... | abstraction_interface | DCAgent/swesmith-sandboxes-with_tests-25k |
swesmith-000002 | [
31,
139,
8,
0,
0,
0,
0,
0,
2,
255,
237,
125,
253,
146,
219,
54,
178,
175,
255,
214,
83,
96,
149,
83,
187,
158,
242,
232,
211,
51,
154,
201,
100,
103,
55,
19,
219,
73,
124,
78,
252,
81,
246,
56,
91,
167,
50,
41,
26,
34,... | side_effects_purity | DCAgent/swesmith-sandboxes-with_tests-25k |
swesmith-000003 | [
31,
139,
8,
0,
0,
0,
0,
0,
2,
255,
237,
125,
251,
122,
227,
184,
177,
231,
252,
173,
167,
64,
52,
39,
99,
59,
109,
209,
150,
124,
29,
231,
104,
50,
238,
182,
167,
219,
233,
235,
113,
187,
103,
206,
126,
211,
253,
81,
20,... | testability_design | DCAgent/swesmith-sandboxes-with_tests-25k |
swesmith-000004 | [
31,
139,
8,
0,
0,
0,
0,
0,
2,
255,
237,
93,
123,
119,
219,
182,
146,
207,
223,
250,
20,
184,
234,
238,
181,
125,
35,
81,
15,
203,
143,
186,
215,
119,
227,
38,
105,
155,
221,
54,
206,
113,
156,
222,
179,
167,
201,
97,
41,... | duplication_redundancy | DCAgent/swesmith-sandboxes-with_tests-25k |
swesmith-000005 | [
31,
139,
8,
0,
0,
0,
0,
0,
2,
255,
237,
125,
253,
118,
219,
56,
178,
103,
254,
214,
83,
96,
212,
247,
78,
219,
39,
22,
45,
201,
182,
236,
246,
140,
231,
182,
59,
73,
79,
188,
183,
59,
201,
73,
156,
153,
179,
167,
157,
... | backwards_compat | DCAgent/swesmith-sandboxes-with_tests-25k |
swesmith-000006 | [
31,
139,
8,
0,
0,
0,
0,
0,
2,
255,
237,
125,
253,
150,
219,
54,
146,
175,
255,
214,
83,
96,
228,
221,
184,
125,
220,
146,
90,
234,
79,
119,
166,
103,
211,
99,
59,
177,
239,
58,
110,
31,
187,
157,
217,
123,
98,
31,
26,
... | testing_observability | DCAgent/swesmith-sandboxes-with_tests-25k |
swesmith-000007 | [
31,
139,
8,
0,
0,
0,
0,
0,
2,
255,
237,
125,
251,
114,
219,
198,
210,
167,
255,
230,
83,
204,
97,
190,
138,
164,
178,
72,
138,
212,
213,
202,
209,
57,
81,
108,
39,
209,
126,
137,
229,
178,
157,
164,
182,
98,
23,
2,
2,
... | performance | DCAgent/swesmith-sandboxes-with_tests-25k |
swesmith-000008 | [
31,
139,
8,
0,
0,
0,
0,
0,
2,
255,
237,
93,
123,
115,
220,
54,
146,
247,
223,
243,
41,
176,
227,
171,
149,
92,
214,
80,
51,
163,
103,
148,
213,
94,
20,
91,
137,
181,
151,
88,
46,
91,
78,
234,
42,
118,
49,
24,
18,
51,
... | flexibility_params | DCAgent/swesmith-sandboxes-with_tests-25k |
swesmith-000009 | [
31,
139,
8,
0,
0,
0,
0,
0,
2,
255,
237,
125,
123,
147,
219,
54,
178,
175,
255,
214,
167,
64,
148,
83,
59,
227,
120,
196,
145,
52,
207,
76,
118,
246,
100,
214,
118,
18,
223,
147,
216,
46,
219,
201,
214,
173,
216,
69,
67,
... | clarity_complexity | DCAgent/swesmith-sandboxes-with_tests-25k |
swesmith-000010 | [
31,
139,
8,
0,
0,
0,
0,
0,
2,
255,
237,
125,
253,
118,
219,
182,
178,
111,
254,
214,
83,
96,
171,
103,
111,
219,
43,
214,
167,
109,
217,
117,
183,
247,
169,
155,
164,
109,
206,
105,
147,
44,
39,
105,
215,
93,
73,
22,
11,... | modularity_reuse | DCAgent/swesmith-sandboxes-with_tests-25k |
swesmith-000011 | [
31,
139,
8,
0,
0,
0,
0,
0,
2,
255,
237,
125,
235,
118,
219,
56,
178,
110,
126,
235,
41,
48,
202,
94,
99,
123,
98,
235,
230,
91,
226,
25,
207,
110,
79,
146,
238,
246,
153,
206,
101,
37,
233,
238,
117,
86,
146,
197,
64,
... | data_transform | DCAgent/swesmith-sandboxes-with_tests-25k |
swesmith-000012 | [
31,
139,
8,
0,
0,
0,
0,
0,
2,
255,
237,
125,
125,
115,
227,
54,
210,
231,
252,
173,
79,
129,
117,
158,
141,
237,
26,
155,
178,
100,
91,
246,
56,
235,
221,
76,
198,
147,
100,
238,
73,
198,
83,
30,
39,
91,
87,
153,
20,
1... | idempotency | DCAgent/swesmith-sandboxes-with_tests-25k |
End of preview. Expand in Data Studio
YAML Metadata Warning:empty or missing yaml metadata in repo card
Check out the documentation for more information.
SERAlike-swesmith-25k
A drop-in replacement for DCAgent/swesmith-sandboxes-with_tests-25k where the per-row instruction.md is rewritten from the original specific SWE-Smith bug report into a SERA-style vague, "patch or abstain" prompt.
Why
SERA (2026) trains agents on synthetic data generated by the Soft Verified Generation (SVG) inversion: the teacher is given a vague bug-type hint, the agent invents a bug-fix, and the model's output is taken as ground truth (the "bug" being whatever the model decided to change). SERA's published Sera-4.6-Lite-T2-v4 corpus is generated from the SWE-Smith codebase set (121 repos, per the paper).
This dataset reuses SWE-Smith's existing real-bug task tarballs (Dockerfile + tests + solution all preserved) but replaces the instruction with a SERA-style vague prompt drawn from one of SERA's 51 ROLLOUT_ONE_PROMPTS. The agent is then explicitly given an abstain branch in the prompt: if it cannot identify a real change, it outputs <abstain/> and stops.
The point: at datagen time we get the vagueness of SERA's prompt distribution while keeping SWE-Smith's real test-based verifier — a strict improvement over SVG's line-level recall, since training only retains traces where the existing tests still pass after the agent's patch.
Schema
| column | type | notes |
|---|---|---|
path |
string | row id (swesmith-XXXXXX), inherited from upstream |
task_binary |
bytes | gzipped tar with the same layout as upstream — instruction.md (rewritten), task.toml, environment/Dockerfile, tests/*, solution/* |
bug_label |
string | which of SERA's 51 bug categories was sampled (e.g. state_management, edge_cases) |
source_dataset |
string | DCAgent/swesmith-sandboxes-with_tests-25k |
The bug label is sampled deterministically from sha256(path) so the rewrite is reproducible.
What changes vs upstream
instruction.md— replaced.- Everything else —
task.toml,environment/Dockerfile,tests/test.sh,tests/test_state.py,tests/config.json,solution/solve.sh— unchanged, including the SWE-Smith Docker image reference and the buggy commit checkout.
This means agents see SERA-style vague prompts but the underlying ground-truth bug + the existing test verifier are SWE-Smith's. Patches that don't fix what's actually broken should still fail at verification time, so post-hoc trace filtering is straightforward.
Generation
python -m data.seralike.rewrite # builds shards into ./out/
python -m data.seralike.upload --repo DCAgent/SERAlike-swesmith-25k
Source: OpenThoughts-Agent/data/seralike/.
- Downloads last month
- 49