Datasets:
Document public v0.1 research snapshot
Browse files- DATA_RIGHTS.md +2 -2
- MANIFEST.sha256.json +3 -2
- README.md +3 -3
- provenance/publication.json +15 -0
DATA_RIGHTS.md
CHANGED
|
@@ -1,7 +1,7 @@
|
|
| 1 |
# 数据来源与权利记录
|
| 2 |
|
| 3 |
-
本版本
|
| 4 |
|
| 5 |
`provenance/generation_cases.jsonl` 保留原始语料的 DOI、定位、label_kind、已记录的 license_kind 和审核状态;`provenance/source_catalog.jsonl` 汇总各子集可识别来源。`license_kind` 是上游记录,未在本次上传中独立复核。缺失的许可标记为待核对。
|
| 6 |
|
| 7 |
-
|
|
|
|
| 1 |
# 数据来源与权利记录
|
| 2 |
|
| 3 |
+
本版本为公开研究快照,供同行查看与复现。包含项目构造的题目、参考答案、已有评分代码,以及第三方论文的原文段落、表格内容和引用。原始论文内容的权利归其权利人,不能因为本仓库上传就推定其适用统一开源许可。
|
| 4 |
|
| 5 |
`provenance/generation_cases.jsonl` 保留原始语料的 DOI、定位、label_kind、已记录的 license_kind 和审核状态;`provenance/source_catalog.jsonl` 汇总各子集可识别来源。`license_kind` 是上游记录,未在本次上传中独立复核。缺失的许可标记为待核对。
|
| 6 |
|
| 7 |
+
各来源许可证原文、署名与再分发条件逐项核对和补充维护中;项目原创题目、标注和评分代码尚未指定统一发布许可。本快照不替权利人授予新的许可,不包含模型权重许可授权。
|
MANIFEST.sha256.json
CHANGED
|
@@ -2,8 +2,8 @@
|
|
| 2 |
"algorithm": "SHA-256",
|
| 3 |
"files": {
|
| 4 |
".gitignore": "b72e95d31b8da405ae3f609d60c88628263853f0016653a8e9f3fe245290ac70",
|
| 5 |
-
"DATA_RIGHTS.md": "
|
| 6 |
-
"README.md": "
|
| 7 |
"audits/package_validation.json": "6e90929899678ae13d3af411524210188180a647fc586e98aeb9bec0c71dc354",
|
| 8 |
"audits/question_reference_audit.json": "4e66e30ca4ce0a467356b1495bb504c755bda1d8927f81fe5161d582711526c8",
|
| 9 |
"configs/analysis_plan.json": "0e4a587f44cdad459f9ff237b867ca2f81cc37477905be6612e5b35591a26618",
|
|
@@ -18,6 +18,7 @@
|
|
| 18 |
"inputs/workflow.jsonl": "e93345326fb081fc281b820b1e77a4c4fd9a0409caccef806deb886fd1fdfaaf",
|
| 19 |
"provenance/generation_cases.jsonl": "eb4e4ecf902bdf8bac0182571f8d582479d9d9a851c74447f069b0b3f5afe4a9",
|
| 20 |
"provenance/generation_clusters.json": "e1a5b359746c3f4b1a1301924d5e85d2005fadd876d218cc88a0804f6b3ab44d",
|
|
|
|
| 21 |
"provenance/release.json": "86a207bc1e1ad93b1d1e15f34eee4f45b909018eb8be537742efd27cb9eef8b6",
|
| 22 |
"provenance/scorer_port.json": "818eaf344827059ed0eec2b3ba8472d95c6f3340fc5f119bde693265dac817e7",
|
| 23 |
"provenance/source_catalog.jsonl": "1ee756b83885c3a1171813456a030432991bc3c2b4e9cdfc67831ea035623350",
|
|
|
|
| 2 |
"algorithm": "SHA-256",
|
| 3 |
"files": {
|
| 4 |
".gitignore": "b72e95d31b8da405ae3f609d60c88628263853f0016653a8e9f3fe245290ac70",
|
| 5 |
+
"DATA_RIGHTS.md": "c8ef24fe4e311403803cd6c6c1d283a6c88c37d2334cb71c27064ca0341e801e",
|
| 6 |
+
"README.md": "7bbf2b86ed7e7e4d2221936426b31c182c18a12b18c210b394caf615c37f9284",
|
| 7 |
"audits/package_validation.json": "6e90929899678ae13d3af411524210188180a647fc586e98aeb9bec0c71dc354",
|
| 8 |
"audits/question_reference_audit.json": "4e66e30ca4ce0a467356b1495bb504c755bda1d8927f81fe5161d582711526c8",
|
| 9 |
"configs/analysis_plan.json": "0e4a587f44cdad459f9ff237b867ca2f81cc37477905be6612e5b35591a26618",
|
|
|
|
| 18 |
"inputs/workflow.jsonl": "e93345326fb081fc281b820b1e77a4c4fd9a0409caccef806deb886fd1fdfaaf",
|
| 19 |
"provenance/generation_cases.jsonl": "eb4e4ecf902bdf8bac0182571f8d582479d9d9a851c74447f069b0b3f5afe4a9",
|
| 20 |
"provenance/generation_clusters.json": "e1a5b359746c3f4b1a1301924d5e85d2005fadd876d218cc88a0804f6b3ab44d",
|
| 21 |
+
"provenance/publication.json": "2acaa28fbb2ed6d199fc71146a01756b8ae69d358f23dbdc10e723c08d3267e8",
|
| 22 |
"provenance/release.json": "86a207bc1e1ad93b1d1e15f34eee4f45b909018eb8be537742efd27cb9eef8b6",
|
| 23 |
"provenance/scorer_port.json": "818eaf344827059ed0eec2b3ba8472d95c6f3340fc5f119bde693265dac817e7",
|
| 24 |
"provenance/source_catalog.jsonl": "1ee756b83885c3a1171813456a030432991bc3c2b4e9cdfc67831ea035623350",
|
README.md
CHANGED
|
@@ -32,7 +32,7 @@ configs:
|
|
| 32 |
|
| 33 |
# PI Research Bench · 聚酰亚胺科研任务评测基准
|
| 34 |
|
| 35 |
-
**v0.1
|
| 36 |
|
| 37 |
| 子集 | 评测单元 | 内容 | 主要评价 |
|
| 38 |
|---|---:|---|---|
|
|
@@ -58,7 +58,7 @@ python -m pip install -r requirements.txt
|
|
| 58 |
python scripts/validate_package.py
|
| 59 |
```
|
| 60 |
|
| 61 |
-
通过 Hugging Face Datasets 读取任一子集
|
| 62 |
|
| 63 |
```python
|
| 64 |
import json
|
|
@@ -117,4 +117,4 @@ python scripts/score.py --suite source --predictions source_predictions.jsonl --
|
|
| 117 |
|
| 118 |
此快照按原样保留历史工具消息中的来源路径,它们属于题目文本,不提供远程文件访问能力。部署时使用自己的路径和服务配置。
|
| 119 |
|
| 120 |
-
本版本
|
|
|
|
| 32 |
|
| 33 |
# PI Research Bench · 聚酰亚胺科研任务评测基准
|
| 34 |
|
| 35 |
+
**v0.1 研究快照,2026-10-21,公开供同行查看与复现。** 面向聚酰亚胺科研助手,覆盖领域生成、结构化工具调用、文献证据溯源和科研工作流。题目与参考答案来自当前 Qwen/Gemma 对齐对照实验的冻结输入。本版本保存题库、机械评分器、工作流审查量表和来源记录。
|
| 36 |
|
| 37 |
| 子集 | 评测单元 | 内容 | 主要评价 |
|
| 38 |
|---|---:|---|---|
|
|
|
|
| 58 |
python scripts/validate_package.py
|
| 59 |
```
|
| 60 |
|
| 61 |
+
通过 Hugging Face Datasets 读取任一子集:
|
| 62 |
|
| 63 |
```python
|
| 64 |
import json
|
|
|
|
| 117 |
|
| 118 |
此快照按原样保留历史工具消息中的来源路径,它们属于题目文本,不提供远程文件访问能力。部署时使用自己的路径和服务配置。
|
| 119 |
|
| 120 |
+
本版本已公开。文献段落及原表内容适用各原始来源的权利与许可,详见 `DATA_RIGHTS.md`;逐来源许可核对和完整题目审查仍在进行。当前运行的模型结果会在统一评分与内容审查完成后另行整理;本版本没有排行榜结论。
|
provenance/publication.json
ADDED
|
@@ -0,0 +1,15 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"utc": "2026-10-21T06:55:18.002300+00:00",
|
| 3 |
+
"visibility": "public",
|
| 4 |
+
"authorization": "Owner requested public access to share the benchmark with research colleagues.",
|
| 5 |
+
"previous_private_commit": "cb288cf890fac88c4331c3605f16e9a0ffd6ce6b",
|
| 6 |
+
"benchmark_version": "0.1",
|
| 7 |
+
"input_reference_and_scoring_content_changed": false,
|
| 8 |
+
"scientific_review_status": "unchanged; pending domain review",
|
| 9 |
+
"changed_files": [
|
| 10 |
+
"README.md",
|
| 11 |
+
"DATA_RIGHTS.md",
|
| 12 |
+
"provenance/publication.json",
|
| 13 |
+
"MANIFEST.sha256.json"
|
| 14 |
+
]
|
| 15 |
+
}
|