---
license: other
size_categories:
- 10K<n<100K
task_categories:
- text-generation
license_name: mixed-license
license_link: LICENSE.md
tags:
- llm-agents
- deployment-time-learning
- continual-learning
configs:
- config_name: alfworld
  data_files: alfworld/alfworld.jsonl
- config_name: banking77
  data_files: banking77/banking77.jsonl
- config_name: bird
  data_files: bird/bird.jsonl
- config_name: cmdl
  data_files: cmdl/cmdl.jsonl
- config_name: ddxplus
  data_files: ddxplus/ddxplus.jsonl
- config_name: lfd
  data_files: lfd/lfd.jsonl
- config_name: mud
  data_files: mud/mud.jsonl
- config_name: rca
  data_files: rca/rca.jsonl
- config_name: scienceworld
  data_files: scienceworld/scienceworld.jsonl
- config_name: sentifin
  data_files: sentifin/sentifin.jsonl
- config_name: spider
  data_files: spider/spider.jsonl
- config_name: 2wiki
  data_files: 2wiki/2wiki.jsonl
- config_name: ehr
  data_files: ehr/ehr.jsonl
---

# DTLBench

<p align="center">
<a href="https://github.com/guosyjlu/CASCADE">
💻 GitHub Repo
</a> |
<a href="https://physionet.org/content/mimic-iv-ext-dtlbench/1.0.0/">
🫀 DTLBench PhysioNet (To be released)
</a> |
<a href="https://arxiv.org/abs/2605.06702">
📄 Paper
</a>
</p>

DTLBench is a benchmark for **deployment-time learning** of large language model agents. It collects diverse task streams spanning medical diagnosis, legal analysis, operational reasoning, financial prediction, text-to-SQL, embodied decision making, tabular reasoning on EHRs, deep search, etc.

The dataset was introduced in the paper: [CASCADE: Case-Based Continual Adaptation for Large Language Models During Deployment](https://arxiv.org/abs/2605.06702).

## Benchmark Overview

- Total tasks: `16` (3 of them will be released through PhysioNet)
- Data format: one JSON object per line (`.jsonl`)
- Primary use case: benchmark streams for deployment-time learning

DTLBench covers three environment styles used in CASCADE:

- `single-turn`: one input, one final answer
- `multi-turn`: sequential interaction with the environment

### Summary statistics of the DTLBench. The maximum steps refer to the maximum number of interaction steps that the environment allows per task.

| **Property**                 | **Domain**          | **Task**                                          | **Dataset**       | **Maximum Steps** | **Number of Samples** |
|------------------------------|---------------------|--------------------------------------------------|-------------------|-------------------|-----------------------|
| **Single-turn**               | **Medical**         | Medical Diagnosis                                | DDXPlus           | 1                 | 3136                  |
|                              |                     | Medication Recommendation                        | MIMIC-IV-MR       | 1                 | 2881                  |
|                              |                     | Medical Specialty Referral                       | MIMIC-IV-MSR      | 1                 | 2115                  |
|                              |                     | Triage Level Prediction                          | MIMIC-IV-TLP      | 1                 | 2200                  |
|                              | **Legal**           | Multi-Defendant Legal Charge Prediction          | MUD               | 1                 | 1740                  |
|                              |                     | Penalty Legal Prediction                         | CMDL              | 1                 | 2080                  |
|                              | **Financial**       | Financial Customer Intent Routing                | Banking77         | 1                 | 5000                  |
|                              |                     | Entity-Aware Financial Sentiment Analysis        | SEntFiN           | 1                 | 2299                  |
|                              | **AIOps**           | AIOps Root Cause Analysis                        | RCA               | 1                 | 2925                  |
|                              |                     | AIOps Log Fault Diagnosis                        | LFD               | 1                 | 3000                  |
|                              | **Coding**          | Text-to-SQL                                      | SPIDER            | 1                 | 2147                  |
|                              |                     | Knowledge-Augmented Text-to-SQL                 | BIRD              | 1                 | 1534                  |
| **Multi-turn, Simulated**     | **Embodied**        | Household Embodied Decision Making               | ALFWorld          | 30                | 2000                  |
|                              |                     | Scientific Embodied Decision Making              | ScienceWorld      | 10-30             | 1857                  |
| **Multi-turn, Real-world**    | **Information Seeking** | Web-based Deep Search                         | 2Wiki             | 5                 | 2500                  |
|                              | **Medical**         | Complex Tabular Reasoning on Electronic Health Records | MIMIC-III     | 5                 | 2500                  |

## Data Format

Each config can be loaded independently from Hugging Face, and each task keeps the fields needed by its original environment. All tasks include a `task` field, which is the main query or observation presented to the agent.


## Load the Dataset

Using `datasets`:

```python
from datasets import load_dataset

# Load one task
ddxplus = load_dataset("guosy/DTLBench", "ddxplus")
print(ddxplus["train"][0])

# Load another task
spider = load_dataset("guosy/DTLBench", "spider")
print(spider["train"][0]["task"])
```

Using `huggingface-cli`:

```bash
huggingface-cli download --repo-type dataset guosy/DTLBench --local-dir ./DTLBench
```

## License

DTLBench is a **mixed-license** collection. Each subdataset follows its own original license, and the benchmark authors do not claim additional rights beyond those licenses.

Please see [LICENSE.md](LICENSE.md) and the per-task `LICENSE` files for details. In particular:

- Some tasks are under permissive licenses such as MIT or Apache-2.0
- Some tasks use CC licenses with attribution or share-alike requirements
- Some tasks have unclear or unknown redistribution terms

You are responsible for ensuring your use complies with the license of each individual subdataset.

## Citation

If you use DTLBench, please consider citing our paper:
```
@misc{guo2026cascadecasebasedcontinualadaptation,
      title={CASCADE: Case-Based Continual Adaptation for Large Language Models During Deployment}, 
      author={Siyuan Guo and Yali Du and Hechang Chen and Yi Chang and Jun Wang},
      year={2026},
      eprint={2605.06702},
      archivePrefix={arXiv},
      primaryClass={cs.AI},
      url={https://arxiv.org/abs/2605.06702}, 
}
```