Datasets:
ArXiv:
License:
Update README.md
Browse files
README.md
CHANGED
@@ -8,7 +8,7 @@ configs:
|
|
8 |
---
|
9 |
# DEVAI dataset
|
10 |
<p align="center" width="100%">
|
11 |
-
<img src="dataset_stats.png" align="center" width="
|
12 |
</p>
|
13 |
|
14 |
**DEVAI** is a benchmark of 55 realistic AI development tasks. It consists of plentiful manual annotations, including a total of 365 hierarchical user requirements.
|
@@ -16,17 +16,19 @@ This dataset enables rich reinforcement signals for better automated AI software
|
|
16 |
|
17 |
Here is an example of our tasks.
|
18 |
<p align="center" width="100%">
|
19 |
-
<img src="task51.png" align="center" width="
|
20 |
</p>
|
21 |
|
22 |
-
We apply three state-of-the-art automatic software development systems to DEVAI, namely MetaGPT, GPT-Piolt, and OpenHands.
|
|
|
|
|
23 |
<p align="center" width="100%">
|
24 |
-
<img src="developer_stats.png" align="center" width="
|
25 |
</p>
|
26 |
|
27 |
We perform a manual evaluation to judge if each requirement is satisfied by the solution provided by the aforementioned systems.
|
28 |
<p align="center" width="100%">
|
29 |
-
<img src="human_evaluation.png" align="center" width="
|
30 |
</p>
|
31 |
|
32 |
An automated evaluation program that could possibly replace manual evaluation can be found at our [Github realse](https://github.com/metauto-ai/Devai).
|
|
|
8 |
---
|
9 |
# DEVAI dataset
|
10 |
<p align="center" width="100%">
|
11 |
+
<img src="dataset_stats.png" align="center" width="84%"/>
|
12 |
</p>
|
13 |
|
14 |
**DEVAI** is a benchmark of 55 realistic AI development tasks. It consists of plentiful manual annotations, including a total of 365 hierarchical user requirements.
|
|
|
16 |
|
17 |
Here is an example of our tasks.
|
18 |
<p align="center" width="100%">
|
19 |
+
<img src="task51.png" align="center" width="90%"/>
|
20 |
</p>
|
21 |
|
22 |
+
We apply three state-of-the-art automatic software development systems to DEVAI, namely MetaGPT, GPT-Piolt, and OpenHands.
|
23 |
+
We suggest expanding the task queries with constraints defined in [constraints.json](https://huggingface.co/datasets/DEVAI-benchmark/DEVAI/blob/main/constraints.json) to guide development systems' behavior and provide auxiliary if needed.
|
24 |
+
The table below shows preliminary statistics results.
|
25 |
<p align="center" width="100%">
|
26 |
+
<img src="developer_stats.png" align="center" width="79%"/>
|
27 |
</p>
|
28 |
|
29 |
We perform a manual evaluation to judge if each requirement is satisfied by the solution provided by the aforementioned systems.
|
30 |
<p align="center" width="100%">
|
31 |
+
<img src="human_evaluation.png" align="center" width="80%"/>
|
32 |
</p>
|
33 |
|
34 |
An automated evaluation program that could possibly replace manual evaluation can be found at our [Github realse](https://github.com/metauto-ai/Devai).
|