support infer with transformers
add use_hf
This commit is contained in:
@@ -1151,6 +1151,9 @@ python3 dots_ocr/parser.py demo/demo_image1.jpg --prompt prompt_ocr
|
||||
python3 dots_ocr/parser.py demo/demo_image1.jpg --prompt prompt_grounding_ocr --bbox 163 241 1536 705
|
||||
|
||||
```
|
||||
**Based on Transformers**, you can parse an image or a pdf file using the same commands above, just add `--use_hf true`.
|
||||
|
||||
> Notice: transformers is slower than vllm, if you want to use demo/* with transformers,just add `use_hf=True` in `DotsOCRParser(..,use_hf=True)`
|
||||
|
||||
<details>
|
||||
<summary><b>Output Results</b></summary>
|
||||
|
||||
Reference in New Issue
Block a user