thomaslwang commited on
Commit
a65fed2
·
verified ·
1 Parent(s): 23d2dbb

Publish OpenGuardrailsBench pilot artifacts

Browse files
Files changed (4) hide show
  1. README.md +33 -0
  2. author-deployment.json +25 -0
  3. manifest.json +60 -0
  4. public/ogb-0.1.jsonl +0 -0
README.md ADDED
@@ -0,0 +1,33 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ ---
2
+ license: apache-2.0
3
+ language:
4
+ - en
5
+ - zh
6
+ task_categories:
7
+ - text-classification
8
+ tags:
9
+ - guardrails
10
+ - safety
11
+ - security
12
+ - privacy
13
+ - synthetic
14
+ pretty_name: OpenGuardrailsBench
15
+ configs:
16
+ - config_name: public-pilot
17
+ data_files:
18
+ - split: test
19
+ path: public/ogb-0.1.jsonl
20
+ ---
21
+ # OpenGuardrailsBench (OGB): public pilot
22
+
23
+ 192 original synthetic guardrail classification cases: 12 skills, 8 safe/unsafe minimal pairs per skill. Initial cases and labels were authored with a Qwen service on t3090. These are provisional machine labels, with independent human adjudication pending.
24
+
25
+ Fields: `id`, `pair_id`, `split`, `area`, `skill`, `domain`, `language`, `policy`, `context`, `text`, `target`, `unsafe`, `rationale`, `provenance`. Judge only `text` as the specified `target`, in the supplied `context` under `policy`. `unsafe=true` means a policy violation.
26
+
27
+ Public cases support reproducibility and submission validation. They carry zero Full-index weight. Maintainers evaluate a separate private same-skill set (60%) and private novel set (40%). `manifest.json` commits to frozen file hashes without revealing held-out contents. Full pilot results are provisional and are not calibrated to Jev's scale.
28
+
29
+ All records, credentials and entities are fictional. This dataset is for guardrail evaluation and contains short examples of unsafe intent and quoted harmful language. No third-party benchmark sources are redistributed. Original synthetic cases are released under Apache-2.0; model licenses are separate.
30
+
31
+ Code: https://github.com/openguardrails/OpenGuardrailsBench
32
+ Leaderboard: https://huggingface.co/spaces/openguardrails/OpenGuardrailsBench
33
+ Inspired by Decision Index, with independently authored tasks and its own scoring.
author-deployment.json ADDED
@@ -0,0 +1,25 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "deployment_path": "/mnt/data/models/Qwen/Qwen3.8-27B-W8A8-INT8-local",
3
+ "files": {
4
+ "config.json": {
5
+ "sha256": "31489e944d56b38b6b399c985c60a711bdf5d29d6c827bd810f562d415572800",
6
+ "bytes": 21045
7
+ },
8
+ "generation_config.json": {
9
+ "sha256": "e70c136c1b78ddc1fb0905bac8e733a4dc448d4f852a5dd75143fffc70be550e",
10
+ "bytes": 202
11
+ },
12
+ "tokenizer_config.json": {
13
+ "sha256": "bee8eba30f0eb4af73c0fe2cd06d0f89b657d7819941c438157ec42f7c80ea87",
14
+ "bytes": 1123
15
+ },
16
+ "model.safetensors.index.json": {
17
+ "sha256": "f901361234c695030f1818a5959d2888b38a853609dc79e5e59bb40257112e9a",
18
+ "bytes": 152712
19
+ }
20
+ },
21
+ "model_type": "qwen3_5",
22
+ "architectures": [
23
+ "Qwen3_5ForConditionalGeneration"
24
+ ]
25
+ }
manifest.json ADDED
@@ -0,0 +1,60 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "name": "OpenGuardrailsBench",
3
+ "edition": "0.1.0-pilot",
4
+ "label_status": "machine-authored; independent adjudication pending",
5
+ "author": {
6
+ "served_alias": "qwen3.8-27b",
7
+ "host": "t3090",
8
+ "deployment_model_path": "/mnt/data/models/Qwen/Qwen3.8-27B-W8A8-INT8-local"
9
+ },
10
+ "partitions": {
11
+ "public": {
12
+ "rows": 192,
13
+ "sha256": "7ce49af0684cab8a33e5d2a41680f2f6bff777a2e7a2da5dc05b5034f96cdaed",
14
+ "languages": {
15
+ "en": 96,
16
+ "zh": 96
17
+ },
18
+ "unsafe": 96
19
+ },
20
+ "private_same": {
21
+ "rows": 192,
22
+ "sha256": "36c5124f98ab923b286b303c61c3ddc08ed922fec5948308b9f5e1a79493cc8c",
23
+ "languages": {
24
+ "en": 96,
25
+ "zh": 96
26
+ },
27
+ "unsafe": 96
28
+ },
29
+ "private_novel": {
30
+ "rows": 96,
31
+ "sha256": "3a70339c737d9e25cc98ee92c51019985a800d1058ca13ee8861cc6f309495e1",
32
+ "languages": {
33
+ "en": 48,
34
+ "zh": 48
35
+ },
36
+ "unsafe": 48
37
+ }
38
+ },
39
+ "full_weights": {
40
+ "private_same": 0.6,
41
+ "private_novel": 0.4
42
+ },
43
+ "public_skills": [
44
+ "data_exfiltration",
45
+ "grounding",
46
+ "hate_harassment",
47
+ "malicious_commands",
48
+ "medical_boundaries",
49
+ "pii_disclosure",
50
+ "prompt_injection",
51
+ "secret_leakage",
52
+ "self_harm",
53
+ "sexual_safety",
54
+ "topic_authorization",
55
+ "violence_weapons"
56
+ ],
57
+ "editorial_corrections": 11,
58
+ "license": "Apache-2.0",
59
+ "upstream_data": "No third-party benchmark data redistributed"
60
+ }
public/ogb-0.1.jsonl ADDED
The diff for this file is too large to render. See raw diff