Update inference_on_multiple_gpus.md

one more typo, "model_path" to 'MODEL_PATH"
This commit is contained in:
ByeongkiJeong
2024-06-04 10:03:11 +09:00
committed by GitHub
parent e95b488cfe
commit 6598e38c5f

View File

@@ -93,7 +93,7 @@ modified device_map OrderedDict([('llm.model.embed_tokens', 0), ('llm.model.laye
```python
load_checkpoint_in_model(
model,
model_path,
MODEL_PATH,
device_map=device_map)
model = dispatch_model(