Add license information
Browse files
README.md
CHANGED
|
@@ -57,7 +57,10 @@ Please also check the raw data after our processing if you are interested: [love
|
|
| 57 |
</table>
|
| 58 |
|
| 59 |
**Introduction**
|
| 60 |
-
|
| 61 |
While having full executable code theoretically allows us to generate reliable execution trajectories as responses, two challenges arise: 1) Obtaining a deterministic reverse function for input prediction is impractical; 2) Automatically constructed trajectories are constrained by pre-designed templates and lack the expressiveness and generalizability of free-form natural language reasoning. Thus, we adopt a fully LLM-based approach for synthesizing all the desired responses using DeepSeek-V2.5, as it has top-tier performance but extremely low cost compared to other advanced LLMs.
|
| 62 |
|
| 63 |
-
*Due to our collaborators' compliance requirements, we only release the PythonEdu-Rs subset (this page) of full dataset.
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 57 |
</table>
|
| 58 |
|
| 59 |
**Introduction**
|
|
|
|
| 60 |
While having full executable code theoretically allows us to generate reliable execution trajectories as responses, two challenges arise: 1) Obtaining a deterministic reverse function for input prediction is impractical; 2) Automatically constructed trajectories are constrained by pre-designed templates and lack the expressiveness and generalizability of free-form natural language reasoning. Thus, we adopt a fully LLM-based approach for synthesizing all the desired responses using DeepSeek-V2.5, as it has top-tier performance but extremely low cost compared to other advanced LLMs.
|
| 61 |
|
| 62 |
+
*Due to our collaborators' compliance requirements, we only release the PythonEdu-Rs subset (this page) of full dataset.
|
| 63 |
+
|
| 64 |
+
**License**
|
| 65 |
+
|
| 66 |
+
The license for this dataset is odc-by.
|