loubnabnl HF staff commited on
Commit
9196092
1 Parent(s): d41708b
Files changed (1) hide show
  1. evaluation/intro.txt +0 -3
evaluation/intro.txt CHANGED
@@ -1,7 +1,6 @@
1
  A popular evaluation framework for code generation models is the [pass@k](https://huggingface.co/metrics/code_eval) metric on [HumanEval](https://huggingface.co/datasets/openai_humaneval) dataset, which was introduced in [Codex paper](https://arxiv.org/pdf/2107.03374v2.pdf). The dataset includes 164 handwritten programming problems. In the pass@k metric, k code samples are generated per problem, a problem is considered solved if any sample passes the unit tests and the total fraction of problems solved is reported.
2
  In most papers, 200 candidate program completions are sampled, and pass@1, pass@10, and pass@100 are computed using an unbiased sampling estimator. The table below shows the HumanEval scores of CodeParrot, InCoder, GPT-neo, GPT-J and Codex (not open-source).
3
 
4
- <div align="center">
5
 
6
  | Model | pass@1 | pass@10 | pass@100|
7
  |-------|--------|---------|---------|
@@ -16,8 +15,6 @@ In most papers, 200 candidate program completions are sampled, and pass@1, pass@
16
  |GPT-neo (1.5B)| 4.79% | 7.47% | 16.30% |
17
  |GPT-J (6B)| 11.62% | 15.74% | 27.74% |
18
 
19
- </div>
20
-
21
 
22
  To better understand how pass@k metric works, we will illustrate it with some examples. We select 4 problems from the HumanEval dataset and see how the model performs and which code completions pass the unit tests. We will use CodeParrot 🦜 with the three problem below:
23
 
 
1
  A popular evaluation framework for code generation models is the [pass@k](https://huggingface.co/metrics/code_eval) metric on [HumanEval](https://huggingface.co/datasets/openai_humaneval) dataset, which was introduced in [Codex paper](https://arxiv.org/pdf/2107.03374v2.pdf). The dataset includes 164 handwritten programming problems. In the pass@k metric, k code samples are generated per problem, a problem is considered solved if any sample passes the unit tests and the total fraction of problems solved is reported.
2
  In most papers, 200 candidate program completions are sampled, and pass@1, pass@10, and pass@100 are computed using an unbiased sampling estimator. The table below shows the HumanEval scores of CodeParrot, InCoder, GPT-neo, GPT-J and Codex (not open-source).
3
 
 
4
 
5
  | Model | pass@1 | pass@10 | pass@100|
6
  |-------|--------|---------|---------|
 
15
  |GPT-neo (1.5B)| 4.79% | 7.47% | 16.30% |
16
  |GPT-J (6B)| 11.62% | 15.74% | 27.74% |
17
 
 
 
18
 
19
  To better understand how pass@k metric works, we will illustrate it with some examples. We select 4 problems from the HumanEval dataset and see how the model performs and which code completions pass the unit tests. We will use CodeParrot 🦜 with the three problem below:
20