Hugging Face's logo

opencsg
/
opencsg-CodeLlama-34b-v0.1

Text Generation
Transformers
Safetensors
code
llama
llama-2
text-generation-inference
Model card Files Community
TomPei commited on
Commit
8119074
·
verified ·
1 Parent(s): 685cda2

Update README.md

Browse files
Files changed (1) hide show
  1. README.md +9 -3
README.md CHANGED
@@ -42,6 +42,8 @@ This is the repository for the base 34B version finetuned based on [CodeLlama-34
42
  | 7B | [opencsg/Opencsg-CodeLlama-7b-v0.1](https://huggingface.co/opencsg/opencsg-CodeLlama-7b-v0.1) |
43
  | 13B | [opencsg/Opencsg-CodeLlama-13b-v0.1](https://huggingface.co/opencsg/opencsg-CodeLlama-13b-v0.1) |
44
  | 34B | [opencsg/Opencsg-CodeLlama-34b-v0.1](https://huggingface.co/opencsg/opencsg-CodeLlama-34b-v0.1) |
 
 
45
 
46
 
47
  ## Model Eval
@@ -60,11 +62,13 @@ To simplify the comparison, we chosed the Pass@1 metric for the Python language,
60
  | Model | HumanEval python pass@1 |
61
  | --- |----------------------------------------------------------------------------- |
62
  | CodeLlama-7b-hf | 30.5%|
63
- | opencsg-CodeLlama-7b-v0.1(4k) | **43.9%** |
64
  | CodeLlama-13b-hf | 36.0%|
65
- | opencsg-CodeLlama-13b-v0.1(4k) | **51.2%** |
66
  | CodeLlama-34b-hf | 48.2%|
67
- | opencsg-CodeLlama-34b-v0.1(4k)| **56.1%** |
 
 
68
 
69
  **TODO**
70
  - We will provide more benchmark scores on fine-tuned models in the future.
@@ -153,6 +157,7 @@ opencsg-CodeLlama-v0.1是一系列基于CodeLlama的通过全参数微调方法
153
  | 7B | [opencsg/Opencsg-CodeLlama-7b-v0.1](https://huggingface.co/opencsg/opencsg-CodeLlama-7b-v0.1) |
154
  | 13B | [opencsg/Opencsg-CodeLlama-13b-v0.1](https://huggingface.co/opencsg/opencsg-CodeLlama-13b-v0.1) |
155
  | 34B | [opencsg/Opencsg-CodeLlama-34b-v0.1](https://huggingface.co/opencsg/opencsg-CodeLlama-34b-v0.1) |
 
156
 
157
 
158
  ## 模型评估
@@ -176,6 +181,7 @@ HumanEval 是评估模型在代码生成方面性能的最常见的基准，尤
176
  | opencsg-CodeLlama-13b-v0.1 | **51.2%** |
177
  | CodeLlama-34b-hf | 48.2%|
178
  | opencsg-CodeLlama-34b-v0.1| **56.1%** |
 
179
 
180
  **TODO**
181
  - 未来我们将提供更多微调模型的在各基准上的分数。
 
42
  | 7B | [opencsg/Opencsg-CodeLlama-7b-v0.1](https://huggingface.co/opencsg/opencsg-CodeLlama-7b-v0.1) |
43
  | 13B | [opencsg/Opencsg-CodeLlama-13b-v0.1](https://huggingface.co/opencsg/opencsg-CodeLlama-13b-v0.1) |
44
  | 34B | [opencsg/Opencsg-CodeLlama-34b-v0.1](https://huggingface.co/opencsg/opencsg-CodeLlama-34b-v0.1) |
45
+ | 34B | [opencsg/Opencsg-CodeLlama-34b-v0.2](https://huggingface.co/opencsg/opencsg-CodeLlama-34b-v0.2) |
46
+
47
 
48
 
49
  ## Model Eval
 
62
  | Model | HumanEval python pass@1 |
63
  | --- |----------------------------------------------------------------------------- |
64
  | CodeLlama-7b-hf | 30.5%|
65
+ | opencsg-CodeLlama-7b-v0.1 | **43.9%** |
66
  | CodeLlama-13b-hf | 36.0%|
67
+ | opencsg-CodeLlama-13b-v0.1 | **51.2%** |
68
  | CodeLlama-34b-hf | 48.2%|
69
+ | opencsg-CodeLlama-34b-v0.1| **56.1%** |
70
+ | opencsg-CodeLlama-34b-v0.2| **64.0%** |
71
+
72
 
73
  **TODO**
74
  - We will provide more benchmark scores on fine-tuned models in the future.
 
157
  | 7B | [opencsg/Opencsg-CodeLlama-7b-v0.1](https://huggingface.co/opencsg/opencsg-CodeLlama-7b-v0.1) |
158
  | 13B | [opencsg/Opencsg-CodeLlama-13b-v0.1](https://huggingface.co/opencsg/opencsg-CodeLlama-13b-v0.1) |
159
  | 34B | [opencsg/Opencsg-CodeLlama-34b-v0.1](https://huggingface.co/opencsg/opencsg-CodeLlama-34b-v0.1) |
160
+ | 34B | [opencsg/Opencsg-CodeLlama-34b-v0.2](https://huggingface.co/opencsg/opencsg-CodeLlama-34b-v0.2) |
161
 
162
 
163
  ## 模型评估
 
181
  | opencsg-CodeLlama-13b-v0.1 | **51.2%** |
182
  | CodeLlama-34b-hf | 48.2%|
183
  | opencsg-CodeLlama-34b-v0.1| **56.1%** |
184
+ | opencsg-CodeLlama-34b-v0.1| **64.0%** |
185
 
186
  **TODO**
187
  - 未来我们将提供更多微调模型的在各基准上的分数。