Commit
•
bbf77de
1
Parent(s):
4acf7b4
Update README.md
Browse files
README.md
CHANGED
|
@@ -30,6 +30,8 @@ Apache-2.0
|
|
| 30 |
|
| 31 |
## Usage
|
| 32 |
|
|
|
|
|
|
|
| 33 |
Loading model and tokenizer:
|
| 34 |
|
| 35 |
```python
|
|
|
|
| 30 |
|
| 31 |
## Usage
|
| 32 |
|
| 33 |
+
At full precision the model needs > 48G GPU memory. A single A100-80GB GPU suffices, for example. If you're running on smaller GPUs, you need an instance with multiple GPUs and/or reduced model precision (e.g. use model.half() before moving to device)
|
| 34 |
+
|
| 35 |
Loading model and tokenizer:
|
| 36 |
|
| 37 |
```python
|