Create README.md
Browse files
README.md
ADDED
|
@@ -0,0 +1,47 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
## Dataset Description
|
| 2 |
+
|
| 3 |
+
This dataset contains the test splits used to evaluate the Memory Decoder model across three specialized domains: biomedical, legal, and finance. The test data was randomly sampled from publicly available datasets to assess the model's performance in domain-specific language understanding.
|
| 4 |
+
|
| 5 |
+
## Dataset Sources
|
| 6 |
+
|
| 7 |
+
The test data is sampled randomly from the following source datasets:
|
| 8 |
+
|
| 9 |
+
### Biomedical Domain
|
| 10 |
+
- **Source:** [Medilora/mimic_iii_diagnosis_anonymous](https://huggingface.co/datasets/Medilora/mimic_iii_diagnosis_anonymous)
|
| 11 |
+
- **Description:** Anonymized clinical discharge summaries from the MIMIC-III database containing medical diagnoses, patient histories, and treatment procedures
|
| 12 |
+
|
| 13 |
+
### Legal Domain
|
| 14 |
+
- **Source:** [clairebarale/AsyLex](https://huggingface.co/datasets/clairebarale/AsyLex)
|
| 15 |
+
- **Description:** Refugee status determination documents from Canada (1996-2022) containing legal judgments and case analyses
|
| 16 |
+
- **Paper Reference:** [ACL 2023 Findings](https://aclanthology.org/2023.findings-acl.187)
|
| 17 |
+
|
| 18 |
+
### Finance Domain
|
| 19 |
+
- **Source:** [jyanimaulik/yahoo_finance_stockmarket_news](https://huggingface.co/datasets/jyanimaulik/yahoo_finance_stockmarket_news)
|
| 20 |
+
- **Description:** Stock market news articles from Yahoo Finance covering company earnings, market analysis, and financial commentary
|
| 21 |
+
|
| 22 |
+
## Dataset Structure
|
| 23 |
+
|
| 24 |
+
### Data Splits
|
| 25 |
+
|
| 26 |
+
The dataset consists of test splits only, randomly sampled from the source datasets:
|
| 27 |
+
|
| 28 |
+
```
|
| 29 |
+
├── biomed/ # Biomedical domain test split
|
| 30 |
+
├── law/ # Legal domain test split
|
| 31 |
+
└── finance/ # Finance domain test split
|
| 32 |
+
```
|
| 33 |
+
|
| 34 |
+
## Citation
|
| 35 |
+
|
| 36 |
+
```bibtex
|
| 37 |
+
@article{cao2025memory,
|
| 38 |
+
title={Memory decoder: A pretrained, plug-and-play memory for large language models},
|
| 39 |
+
author={Cao, Jiaqi and Wang, Jiarui and Wei, Rubin and Guo, Qipeng and Chen, Kai and Zhou, Bowen and Lin, Zhouhan},
|
| 40 |
+
journal={arXiv preprint arXiv:2508.09874},
|
| 41 |
+
year={2025}
|
| 42 |
+
}
|
| 43 |
+
```
|
| 44 |
+
|
| 45 |
+
## Contact
|
| 46 |
+
|
| 47 |
+
For questions and support: maximus.cao@outlook.com
|