Commit
•
e3582e2
1
Parent(s):
0c46da7
Update README.md
Browse files
README.md
CHANGED
|
@@ -32,7 +32,9 @@ This model was successfully tested on CUDA driver v12.1 and toolkit v11.7 with P
|
|
| 32 |
git clone https://github.com/mit-han-lab/llm-awq \
|
| 33 |
&& cd llm-awq \
|
| 34 |
&& git checkout 71d8e68df78de6c0c817b029a568c064bf22132d \
|
| 35 |
-
&& pip install -e .
|
|
|
|
|
|
|
| 36 |
```
|
| 37 |
|
| 38 |
```python
|
|
@@ -81,7 +83,7 @@ output = model.generate(
|
|
| 81 |
repetition_penalty=1.1,
|
| 82 |
eos_token_id=tokenizer.eos_token_id
|
| 83 |
)
|
| 84 |
-
print(tokenizer.decode(output[0]))
|
| 85 |
```
|
| 86 |
|
| 87 |
## Evaluation
|
|
|
|
| 32 |
git clone https://github.com/mit-han-lab/llm-awq \
|
| 33 |
&& cd llm-awq \
|
| 34 |
&& git checkout 71d8e68df78de6c0c817b029a568c064bf22132d \
|
| 35 |
+
&& pip install -e . \
|
| 36 |
+
&& cd awq/kernels \
|
| 37 |
+
&& python setup.py install
|
| 38 |
```
|
| 39 |
|
| 40 |
```python
|
|
|
|
| 83 |
repetition_penalty=1.1,
|
| 84 |
eos_token_id=tokenizer.eos_token_id
|
| 85 |
)
|
| 86 |
+
print(tokenizer.decode(output[0], skip_special_tokens=True))
|
| 87 |
```
|
| 88 |
|
| 89 |
## Evaluation
|