### Load MemoryArena Datasets with Hugging Face

Source: https://memoryarena.github.io/

Use the Hugging Face Datasets library to load specific subsets of the MemoryArena dataset. Each call loads a different task type.

```python
from datasets import load_dataset

ds = load_dataset("ZexueHe/memoryarena", "bundled_shopping")
ds = load_dataset("ZexueHe/memoryarena", "progressive_search")
ds = load_dataset("ZexueHe/memoryarena", "group_travel_planner")
ds = load_dataset("ZexueHe/memoryarena", "formal_reasoning_math")
ds = load_dataset("ZexueHe/memoryarena", "formal_reasoning_phys")
```

--------------------------------

### Formal Reasoning (Math and Phys) Task Structure

Source: https://memoryarena.github.io/

Represents a task for formal reasoning in mathematics and physics, including background context, questions, and answers.

```json
{
  "id": 0,
  "paper_name": "paper_id",
  "backgrounds": [
    "necessary definitions, formulations, and relevant context of subtask 1",
    "necessary definitions, formulations, and relevant context of subtask 2",
    ...
  ],
  "questions": [
    "Math subtask question 1",
    "Math subtask question 2",
    ...
  ],
  "answers": [
    "Math result for subtask 1",
    "Math result for subtask 2",
    ...
  ]
}
```

--------------------------------

### Bundled Webshop Task Structure

Source: https://memoryarena.github.io/

Represents a task for a bundled webshop, including questions and their corresponding answers.

```json
{
  "id": 0,
  "questions": [
    "search subtask 1",
    "search subtask 1",
    ...
  ],
  "answers": [
    "search subtask result 1",
    "search subtask result 2",
    ...
  ]
}
```

--------------------------------

### Group Travel Planner Task Structure

Source: https://memoryarena.github.io/

Illustrates a complex task for planning group travel, including base person details and multiple participant queries.

```json
{
  "id": 0,
  "base_person": {
    "name": "Jennifer",
    "query": "I am Jennifer. Please help me plan a trip from St. Petersburg to Rockford spanning 3 days...",
    "daily_plans": [
      {"days": 1, "current_city": "from St. Petersburg to Rockford", "transportation": "..."},
      {"days": 2, "current_city": "Rockford", "transportation": "..."},
      ...
    ]
  },
  "questions": [
    "I am Eric.\n I'm joining Jennifer for this trip...",
    "I am Emma.\n I'm traveling with Jennifer and Eric...",
    ...
  ],
  "answers": [
    [
      {"days": 1, "current_city": "from St. Petersburg to Rockford", "transportation": "..."},
      {"days": 2, "current_city": "Rockford", "transportation": "..."},
      ...
    ],
    [
      {"days": 1, "current_city": "from St. Petersburg to Rockford", "transportation": "..."},
      {"days": 2, "current_city": "Rockford", "transportation": "..."},
      ...
    ],
    ...
  ]
}
```

--------------------------------

### Progressive Search Task Structure

Source: https://memoryarena.github.io/

Defines a task for progressive search, where questions lead to specific product targets and attributes.

```json
{
  "id": 0,
  "questions": [
    "buy subtask item1",
    "buy subtask item 2",
    ...
  ],
  "answers": [
    {"target_asin": "B00TUDFEW2", "attributes": ["Almond Flour", ...)},
    {"target_asin": "B08957C9ZH", "attributes": [...]},
    ...
  ]
}
```

--------------------------------

### MemoryArena BibTeX Citation

Source: https://memoryarena.github.io/

The BibTeX entry for citing the MemoryArena dataset in academic work.

```bibtex
@article{he2026memoryarena,
  title={MemoryArena: Benchmarking Agent Memory in Interdependent Multi-Session Agentic Tasks},
  author={He, Zexue and Wang, Yu and Zhi, Churan and Hu, Yuanzhe and Chen, Tzu-Ping and Yin, Lang and Chen, Ze and Wu, Tong Arthur and Ouyang, Siru and Wang, Zihan and others},
  journal={arXiv preprint arXiv:2602.16313},
  year={2026}
}
```

=== COMPLETE CONTENT === This response contains all available snippets from this library. No additional content exists. Do not make further requests.