SUMIN/HuggingFace_transformer
1
0
Fork 0
Code Issues Pull Requests Actions 7 Packages Projects Releases Wiki Activity

[docs] use device-agnostic instead of cuda (#35047)

Browse Source 
* fix on xpu

* [run_all]

* add the missing import for Image lib

* add more devices in comment

* bug fix

* replace cuda
This commit is contained in:
Fanli Lin
2024-12-04 02:53:45 +08:00
committed by GitHub
parent 346597b644
commit b8cdc262d5
1 changed files with 13 additions and 7 deletions
Download Patch File Download Diff File Expand all files Collapse all files

20
docs/source/en/perf_torch_compile.md
View File

@@ -27,7 +27,7 @@ To compile any computer vision model of your choice, call `torch.compile()` on t
```diff ```diff
from transformers import AutoModelForImageClassification from transformers import AutoModelForImageClassification
model = AutoModelForImageClassification.from_pretrained(MODEL_ID).to("cuda") model = AutoModelForImageClassification.from_pretrained(MODEL_ID).to(DEVICE)
+ model = torch.compile(model) + model = torch.compile(model)
``` ```
@@ -47,15 +47,17 @@ from PIL import Image
import requests import requests
import numpy as np import numpy as np
from transformers import AutoImageProcessor, AutoModelForImageClassification from transformers import AutoImageProcessor, AutoModelForImageClassification
from accelerate.test_utils.testing import get_backend
device, _, _ = get_backend() # automatically detects the underlying device type (CUDA, CPU, XPU, MPS, etc.)
url = 'http://images.cocodataset.org/val2017/000000039769.jpg' url = 'http://images.cocodataset.org/val2017/000000039769.jpg'
image = Image.open(requests.get(url, stream=True).raw) image = Image.open(requests.get(url, stream=True).raw)
processor = AutoImageProcessor.from_pretrained("google/vit-base-patch16-224") processor = AutoImageProcessor.from_pretrained("google/vit-base-patch16-224")
model = AutoModelForImageClassification.from_pretrained("google/vit-base-patch16-224").to("cuda") model = AutoModelForImageClassification.from_pretrained("google/vit-base-patch16-224").to(device)
model = torch.compile(model) model = torch.compile(model)
processed_input = processor(image, return_tensors='pt').to(device="cuda") processed_input = processor(image, return_tensors='pt').to(device)
with torch.no_grad(): with torch.no_grad():
_ = model(**processed_input) _ = model(**processed_input)
@@ -66,13 +68,15 @@ with torch.no_grad():
```python ```python
from transformers import AutoImageProcessor, AutoModelForObjectDetection from transformers import AutoImageProcessor, AutoModelForObjectDetection
from accelerate.test_utils.testing import get_backend
device, _, _ = get_backend() # automatically detects the underlying device type (CUDA, CPU, XPU, MPS, etc.)
processor = AutoImageProcessor.from_pretrained("facebook/detr-resnet-50") processor = AutoImageProcessor.from_pretrained("facebook/detr-resnet-50")
model = AutoModelForObjectDetection.from_pretrained("facebook/detr-resnet-50").to("cuda") model = AutoModelForObjectDetection.from_pretrained("facebook/detr-resnet-50").to(device)
model = torch.compile(model) model = torch.compile(model)
texts = ["a photo of a cat", "a photo of a dog"] texts = ["a photo of a cat", "a photo of a dog"]
inputs = processor(text=texts, images=image, return_tensors="pt").to("cuda") inputs = processor(text=texts, images=image, return_tensors="pt").to(device)
with torch.no_grad(): with torch.no_grad():
_ = model(**inputs) _ = model(**inputs)
@@ -82,11 +86,13 @@ with torch.no_grad():
```python ```python
from transformers import SegformerImageProcessor, SegformerForSemanticSegmentation from transformers import SegformerImageProcessor, SegformerForSemanticSegmentation
from accelerate.test_utils.testing import get_backend
device, _, _ = get_backend() # automatically detects the underlying device type (CUDA, CPU, XPU, MPS, etc.)
processor = SegformerImageProcessor.from_pretrained("nvidia/segformer-b0-finetuned-ade-512-512") processor = SegformerImageProcessor.from_pretrained("nvidia/segformer-b0-finetuned-ade-512-512")
model = SegformerForSemanticSegmentation.from_pretrained("nvidia/segformer-b0-finetuned-ade-512-512").to("cuda") model = SegformerForSemanticSegmentation.from_pretrained("nvidia/segformer-b0-finetuned-ade-512-512").to(device)
model = torch.compile(model) model = torch.compile(model)
seg_inputs = processor(images=image, return_tensors="pt").to("cuda") seg_inputs = processor(images=image, return_tensors="pt").to(device)
with torch.no_grad(): with torch.no_grad():
_ = model(**seg_inputs) _ = model(**seg_inputs)