Upload dataset with hf_xet
Browse files
README.md
CHANGED
|
@@ -28,30 +28,19 @@ Given SysML v2 code and relevant context (compiler error, or relevant domain rul
|
|
| 28 |
## Fields
|
| 29 |
|
| 30 |
- `id`: Unique identifier for each dataset instance.
|
| 31 |
-
|
| 32 |
- `source_id`: Identifier of the original (clean) code example from which this instance was derived.
|
| 33 |
-
|
| 34 |
- `mutation_category`: High-level category of the applied mutation:
|
| 35 |
- `domain`: domain-specific semantic changes
|
| 36 |
- `syntax`: syntactic errors
|
| 37 |
- `none`: no mutation applied
|
| 38 |
-
|
| 39 |
- `mutation_type`: Specific mutation operator used to generate the erroneous code.
|
| 40 |
-
|
| 41 |
- `bad_code`: SysML v2 code potentially containing injected errors.
|
| 42 |
-
|
| 43 |
- `good_code`: Correct version of the code.
|
| 44 |
-
|
| 45 |
- `diff_patch`: Unified diff representing the transformation from `bad_code` to `good_code`.
|
| 46 |
-
|
| 47 |
- `base_prompt`: Base prompt template used for fine-tuning; contains `bad_code` and compiler error in case of `syntax` mutations.
|
| 48 |
-
|
| 49 |
- `prompt`: Prompt containing additional context (e.g., relevant domain rules), for `domain` and `none` mutations, where no compiler error occurs.
|
| 50 |
-
|
| 51 |
- `code_response`: Repair in full code form.
|
| 52 |
-
|
| 53 |
- `patch_response`: Repair in diff/patch format.
|
| 54 |
-
|
| 55 |
- `length`: Total number of tokens in the full training sequence (prompt + response).
|
| 56 |
|
| 57 |
## Data Creation
|
|
@@ -74,7 +63,6 @@ This dataset is released under CC BY 4.0. Attribution to the original authors is
|
|
| 74 |
## Citation
|
| 75 |
|
| 76 |
GitHub Repository: [SysMLv2 Repair with KG-SLMs](https://github.com/rohailamalik/SysMLv2-repair-with-KG-SLMs)
|
| 77 |
-
|
| 78 |
```bibtex
|
| 79 |
@inproceedings{alshami2026sysml,
|
| 80 |
title={Automated Semantic Fault Localization in SysML v2: A Human-in-the-Loop Framework Using Knowledge-Graph Augmented LLMs},
|
|
|
|
| 28 |
## Fields
|
| 29 |
|
| 30 |
- `id`: Unique identifier for each dataset instance.
|
|
|
|
| 31 |
- `source_id`: Identifier of the original (clean) code example from which this instance was derived.
|
|
|
|
| 32 |
- `mutation_category`: High-level category of the applied mutation:
|
| 33 |
- `domain`: domain-specific semantic changes
|
| 34 |
- `syntax`: syntactic errors
|
| 35 |
- `none`: no mutation applied
|
|
|
|
| 36 |
- `mutation_type`: Specific mutation operator used to generate the erroneous code.
|
|
|
|
| 37 |
- `bad_code`: SysML v2 code potentially containing injected errors.
|
|
|
|
| 38 |
- `good_code`: Correct version of the code.
|
|
|
|
| 39 |
- `diff_patch`: Unified diff representing the transformation from `bad_code` to `good_code`.
|
|
|
|
| 40 |
- `base_prompt`: Base prompt template used for fine-tuning; contains `bad_code` and compiler error in case of `syntax` mutations.
|
|
|
|
| 41 |
- `prompt`: Prompt containing additional context (e.g., relevant domain rules), for `domain` and `none` mutations, where no compiler error occurs.
|
|
|
|
| 42 |
- `code_response`: Repair in full code form.
|
|
|
|
| 43 |
- `patch_response`: Repair in diff/patch format.
|
|
|
|
| 44 |
- `length`: Total number of tokens in the full training sequence (prompt + response).
|
| 45 |
|
| 46 |
## Data Creation
|
|
|
|
| 63 |
## Citation
|
| 64 |
|
| 65 |
GitHub Repository: [SysMLv2 Repair with KG-SLMs](https://github.com/rohailamalik/SysMLv2-repair-with-KG-SLMs)
|
|
|
|
| 66 |
```bibtex
|
| 67 |
@inproceedings{alshami2026sysml,
|
| 68 |
title={Automated Semantic Fault Localization in SysML v2: A Human-in-the-Loop Framework Using Knowledge-Graph Augmented LLMs},
|