File size: 2,042 Bytes
9d19e21
d2385cc
9d19e21
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
d2385cc
9d19e21
 
 
 
 
 
 
 
 
 
 
 
d2385cc
 
 
 
 
 
 
 
 
 
 
 
 
 
3993559
 
 
 
d2385cc
 
 
 
3993559
d2385cc
3993559
 
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
---
license: cc-by-4.0
task_categories:
- reinforcement-learning
language:
- en
tags:
- recursive-task-synthesis
- agent-trajectories
- command-line
configs:
- config_name: default
  data_files:
  - split: train
    path: metadata/trajectories.parquet
---

# Recursive Task Synthesis Trajectories

This dataset contains 327,189 completed agent trajectories collected on
recursively synthesized command-line tasks. Public identifiers are opaque and
stable.

The trajectory JSON retains messages, actions, observations, and token counts.
Token-level log-probability arrays and duplicated debug/session captures are
excluded from the public packages.

- `metadata/trajectories.parquet`: searchable trajectory metadata.
- `metadata/shard_manifest.jsonl`: TAR sizes and SHA256 checksums.
- `data/trajectories-*.tar`: sanitized trajectory packages.

```python
from datasets import load_dataset

trajectories = load_dataset(
    "Zhongzhi1228/Recursive-Task-Synthesis-Trajectories",
    split="train",
)
```

## License, attribution, and modifications

This dataset is released under the
[Creative Commons Attribution 4.0 International License](https://creativecommons.org/licenses/by/4.0/).
See `LICENSE` for the complete legal text.

These trajectories were generated by agents interacting with tasks from the
[Recursive Task Synthesis dataset](https://huggingface.co/datasets/Zhongzhi1228/Recursive-Task-Synthesis).
They capture complete agent interactions across validated command-line
environments, including messages, actions, observations, execution outcomes,
rewards, and token statistics. The records were collected, filtered, sanitized,
assigned stable public identifiers, and repackaged for analysis and training.

This license covers the original contributions and
adaptations that the publisher has authority to license. It does not grant
rights over third-party materials beyond what is permitted by their original
terms.

Detailed methodology, data construction, and related-work attribution are
documented in the accompanying paper.