Update app.py
Browse files
app.py
CHANGED
|
@@ -41,6 +41,8 @@ st.sidebar.markdown("""
|
|
| 41 |
|
| 42 |
This app uses the **SmolDocling** model (`ds4sd/SmolDocling-256M-preview`) to convert PDF pages into markdown text. Currently, the model is running on a CPU-based environment (**CPU basic | 2 vCPU - 16 GB RAM**), and therefore processing each page can take a significant amount of time (approximately **6 minutes per page**).
|
| 43 |
|
|
|
|
|
|
|
| 44 |
This setup is suitable for testing and demonstration purposes, but **not efficient for real-world usage**.
|
| 45 |
|
| 46 |
For faster processing, consider running the optimized version `ds4sd/SmolDocling-256M-preview-mlx-bf16` locally on a MacBook, where it performs significantly faster.
|
|
|
|
| 41 |
|
| 42 |
This app uses the **SmolDocling** model (`ds4sd/SmolDocling-256M-preview`) to convert PDF pages into markdown text. Currently, the model is running on a CPU-based environment (**CPU basic | 2 vCPU - 16 GB RAM**), and therefore processing each page can take a significant amount of time (approximately **6 minutes per page**).
|
| 43 |
|
| 44 |
+
**Note: It is recommended that you upload single-page PDFs, as testing showed approximately 6 minutes of processing time per page.**
|
| 45 |
+
|
| 46 |
This setup is suitable for testing and demonstration purposes, but **not efficient for real-world usage**.
|
| 47 |
|
| 48 |
For faster processing, consider running the optimized version `ds4sd/SmolDocling-256M-preview-mlx-bf16` locally on a MacBook, where it performs significantly faster.
|