Update README.md
Browse files
README.md
CHANGED
|
@@ -61,6 +61,10 @@ prompt = "Human: \n" + query + "\n\nAssistant: \n"
|
|
| 61 |
| max new token | 512 |
|
| 62 |
| min new token | 1 |
|
| 63 |
|
|
|
|
|
|
|
|
|
|
|
|
|
| 64 |
其余推理流程与LLaMA2一致
|
| 65 |
|
| 66 |
# 微调
|
|
|
|
| 61 |
| max new token | 512 |
|
| 62 |
| min new token | 1 |
|
| 63 |
|
| 64 |
+
- 量化：
|
| 65 |
+
|
| 66 |
+
在`model.generate()`中添加参数`load_in_8bit=True`或`load_in_4bit=True`
|
| 67 |
+
|
| 68 |
其余推理流程与LLaMA2一致
|
| 69 |
|
| 70 |
# 微调
|