Datasets:

LG-AI-Research
/

DEER-Deep-Research-Benchmark

@@ -1,14 +1,14 @@
-# DEER Dataset License
 Copyright © 2026 LG AI Research. All rights reserved.
-## 1. Purpose
 The DEER dataset is distributed to support non-commercial research, model development, training, and evaluation while preserving the integrity of the benchmark.
 The redistribution and public-indexing restrictions in this license are intended to prevent evaluation contamination, including cases in which web-enabled models or agents retrieve benchmark questions, rubrics, reference materials, or related evaluation content from publicly searchable sources during evaluation.
-## 2. Permitted Uses
 Permission is granted to download and use the DEER dataset solely for non-commercial purposes, including:
@@ -21,7 +21,7 @@ Models, model weights, research results, analyses, and other outputs produced us
 Private storage, processing, and internal sharing within the same organization are permitted solely for the non-commercial purposes described above.
-## 3. Commercial Use Restrictions
 Commercial use of the DEER dataset is prohibited without prior written permission from LG AI Research.
@@ -34,7 +34,7 @@ Commercial use includes, but is not limited to:
 For commercial licensing inquiries, please contact LG AI Research.
-## 4. Redistribution Restrictions
 Without prior written permission from LG AI Research, you may not:
@@ -45,7 +45,7 @@ Without prior written permission from LG AI Research, you may not:
 Users must direct others to the official DEER repository rather than distributing copies of the dataset.
-## 5. Public Search and Indexing Restrictions
 You may not make the original benchmark questions, rubrics, reference materials, or other evaluation content publicly searchable, indexable, browsable, or retrievable through:
@@ -61,17 +61,16 @@ You may not intentionally expose benchmark content in a manner that allows a web
 These restrictions do not prohibit private processing of the dataset for permitted non-commercial research, training, development, or evaluation purposes.
-## 6. Evaluation Integrity
 When evaluating a model or agent on DEER, users should take reasonable measures to prevent the evaluated system from accessing benchmark content through local files, retrieval indexes, cached data, public mirrors, or other unintended sources.
 Evaluation results must not be represented as valid DEER results if the evaluated system had access to the relevant benchmark questions, rubrics, reference materials, or equivalent content before or during evaluation, except where such access is explicitly part of the reported evaluation setting.
-## 7. Citation
 Any publication, released model, technical report, dataset, or other public output that materially uses the DEER dataset must cite the following paper:
-```bibtex
 @inproceedings{
 han2026deer,
 title={{DEER}: A Benchmark for Evaluating Deep Research Agents on Expert Report Generation},
@@ -80,15 +79,14 @@ booktitle={Forty-third International Conference on Machine Learning},
 year={2026},
 url={https://openreview.net/forum?id=ILRx5neJk6}
 }
-```
-## 8. No Warranty
 The dataset is provided “as is,” without warranties or conditions of any kind, express or implied.
 LG AI Research shall not be liable for any claims, damages, losses, or other liabilities arising from the use of the dataset.
-## 9. Reservation of Rights
 No ownership rights are transferred under this license.

+DEER DATASET LICENSE
 Copyright © 2026 LG AI Research. All rights reserved.
+1. PURPOSE
 The DEER dataset is distributed to support non-commercial research, model development, training, and evaluation while preserving the integrity of the benchmark.
 The redistribution and public-indexing restrictions in this license are intended to prevent evaluation contamination, including cases in which web-enabled models or agents retrieve benchmark questions, rubrics, reference materials, or related evaluation content from publicly searchable sources during evaluation.
+2. PERMITTED USES
 Permission is granted to download and use the DEER dataset solely for non-commercial purposes, including:
 Private storage, processing, and internal sharing within the same organization are permitted solely for the non-commercial purposes described above.
+3. COMMERCIAL USE RESTRICTIONS
 Commercial use of the DEER dataset is prohibited without prior written permission from LG AI Research.
 For commercial licensing inquiries, please contact LG AI Research.
+4. REDISTRIBUTION RESTRICTIONS
 Without prior written permission from LG AI Research, you may not:
 Users must direct others to the official DEER repository rather than distributing copies of the dataset.
+5. PUBLIC SEARCH AND INDEXING RESTRICTIONS
 You may not make the original benchmark questions, rubrics, reference materials, or other evaluation content publicly searchable, indexable, browsable, or retrievable through:
 These restrictions do not prohibit private processing of the dataset for permitted non-commercial research, training, development, or evaluation purposes.
+6. EVALUATION INTEGRITY
 When evaluating a model or agent on DEER, users should take reasonable measures to prevent the evaluated system from accessing benchmark content through local files, retrieval indexes, cached data, public mirrors, or other unintended sources.
 Evaluation results must not be represented as valid DEER results if the evaluated system had access to the relevant benchmark questions, rubrics, reference materials, or equivalent content before or during evaluation, except where such access is explicitly part of the reported evaluation setting.
+7. CITATION
 Any publication, released model, technical report, dataset, or other public output that materially uses the DEER dataset must cite the following paper:
 @inproceedings{
 han2026deer,
 title={{DEER}: A Benchmark for Evaluating Deep Research Agents on Expert Report Generation},
 year={2026},
 url={https://openreview.net/forum?id=ILRx5neJk6}
 }
+8. NO WARRANTY
 The dataset is provided “as is,” without warranties or conditions of any kind, express or implied.
 LG AI Research shall not be liable for any claims, damages, losses, or other liabilities arising from the use of the dataset.
+9. RESERVATION OF RIGHTS
 No ownership rights are transferred under this license.