Datasets:

Modalities:
Text
Formats:
parquet
Size:
< 1K
ArXiv:
Libraries:
Datasets
pandas
License:
jiang719 commited on
Commit
f58ccd1
1 Parent(s): ead3abd

Update README.md

Browse files
Files changed (1) hide show
  1. README.md +5 -7
README.md CHANGED
@@ -34,7 +34,11 @@ Large language models (LLMs) have achieved high accuracy, i.e., more than 90 pas
34
 
35
  To address these challenges, we create REPOCOD, a code generation benchmark with 980 problems collected from 11 popular real-world projects, with more than 58% of them requiring file-level or repository-level context information. In addition, REPOCOD has the longest average canonical solution length (331.6 tokens) and the highest average cyclomatic complexity (9.00) compared to existing benchmarks. Each task in REPOCOD includes 313.5 developer-written test cases on average for better correctness evaluation. In our evaluations on ten LLMs, none of the models achieves more than 30 pass@1 on REPOCOD, disclosing the necessity of building stronger LLMs that can help developers in real-world software development.
36
 
37
- For more details on data collection and evaluation results, please refer to our arxiv [preprint](https://arxiv.org/abs/2410.21647).
 
 
 
 
38
 
39
  ## Usage
40
 
@@ -86,12 +90,6 @@ DatasetDict({
86
 
87
  ```
88
 
89
- ## Evaluation Scripts
90
- Examples code for downloading repositories, preparing repository snapshot, and running test cases for evaluation are propived at [code](https://github.com/lt-asset/REPOCOD)
91
-
92
- Check our [Leaderboard](https://lt-asset.github.io/REPOCOD/) for preliminary results using GPT-4o with BM25 and dense retrieval.
93
-
94
-
95
  ## Citation
96
  ```
97
  @misc{liang2024repocod,
 
34
 
35
  To address these challenges, we create REPOCOD, a code generation benchmark with 980 problems collected from 11 popular real-world projects, with more than 58% of them requiring file-level or repository-level context information. In addition, REPOCOD has the longest average canonical solution length (331.6 tokens) and the highest average cyclomatic complexity (9.00) compared to existing benchmarks. Each task in REPOCOD includes 313.5 developer-written test cases on average for better correctness evaluation. In our evaluations on ten LLMs, none of the models achieves more than 30 pass@1 on REPOCOD, disclosing the necessity of building stronger LLMs that can help developers in real-world software development.
36
 
37
+ * For more details on data collection and evaluation results, please refer to our arxiv [preprint](https://arxiv.org/abs/2410.21647).
38
+
39
+ * Examples code for downloading repositories, preparing repository snapshot, and running test cases for evaluation are propived at [code](https://github.com/lt-asset/REPOCOD)
40
+
41
+ * Check our [Leaderboard](https://lt-asset.github.io/REPOCOD/) for preliminary results using GPT-4o with BM25 and dense retrieval.
42
 
43
  ## Usage
44
 
 
90
 
91
  ```
92
 
 
 
 
 
 
 
93
  ## Citation
94
  ```
95
  @misc{liang2024repocod,