@@ -98,7 +98,7 @@ Below you can find the list of the models we benchmarked.
|
||||
- [google/vit-base-patch16-224](https://huggingface.co/google/vit-base-patch16-224)
|
||||
- [microsoft/beit-base-patch16-224-pt22k-ft22k](https://huggingface.co/microsoft/beit-base-patch16-224-pt22k-ft22k)
|
||||
- [facebook/convnext-large-224](https://huggingface.co/facebook/convnext-large-224)
|
||||
- [microsoft/resnet-50](https://huggingface.co/)
|
||||
- [microsoft/resnet-50](https://huggingface.co/microsoft/resnet-50)
|
||||
|
||||
**Image Segmentation**
|
||||
- [nvidia/segformer-b0-finetuned-ade-512-512](https://huggingface.co/nvidia/segformer-b0-finetuned-ade-512-512)
|
||||
|
||||
@@ -157,7 +157,7 @@ Execution time -- 79.0 ms
|
||||
|
||||
Execution time -- 78.9 ms
|
||||
```
|
||||
The first call to `xla_generate()` is time-consuming because of tracing, but the successive calls are orders of magnitude faster. Keep in mind that any change in the generation options at any point with trigger re-tracing and thus leading to slow-downs in the generation time.
|
||||
The first call to `xla_generate()` is time-consuming because of tracing, but the successive calls are orders of magnitude faster. Keep in mind that any change in the generation options at any point will trigger re-tracing and thus leading to slow-downs in the generation time.
|
||||
|
||||
We didn’t cover all the text generation options 🤗 Transformers provides in this document. We encourage you to read the documentation for advanced use cases.
|
||||
|
||||
|
||||
Reference in New Issue
Block a user