yangapku commited on
Commit
1a1d3bf
1 Parent(s): 25d6668

update readme

Browse files
Files changed (1) hide show
  1. README.md +2 -2
README.md CHANGED
@@ -156,9 +156,9 @@ response, history = model.chat(tokenizer, "你好", history=None)
156
 
157
  ### 效果评测
158
 
159
- 我们对BF16和Int4模型在基准评测上做了测试,发现量化模型效果损失较小,结果如下所示:
160
 
161
- We illustrate the model performance of both BF16 and Int4 models on the benchmark, and we find that the quantized model does not suffer from significant performance degradation. Results are shown below:
162
 
163
  | Quantization | MMLU | CEval (val) | GSM8K | Humaneval |
164
  | ------------- | :--------: | :----------: | :----: | :--------: |
 
156
 
157
  ### 效果评测
158
 
159
+ 我们对BF16和Int4模型在基准评测上做了测试(使用zero-shot设置),发现量化模型效果损失较小,结果如下所示:
160
 
161
+ We illustrate the zero-shot performance of both BF16 and Int4 models on the benchmark, and we find that the quantized model does not suffer from significant performance degradation. Results are shown below:
162
 
163
  | Quantization | MMLU | CEval (val) | GSM8K | Humaneval |
164
  | ------------- | :--------: | :----------: | :----: | :--------: |