Updated requirements

Dockerfile

@@ -29,7 +29,7 @@ RUN if [ "$INSTALL_PADDLEOCR" = "True" ]; then \
 fi
 
 RUN if [ "$INSTALL_VLM" = "True" ]; then \
-    pip install --verbose --no-cache-dir --target=/install torch==2.6.0 torchvision --index-url https://download.pytorch.org/whl/cu126; \
+    pip install --verbose --no-cache-dir --target=/install torch==2.8.0 torchvision==0.24.0 --index-url https://download.pytorch.org/whl/cu126; \
     pip install --verbose --no-cache-dir --target=/install transformers==4.57.1 accelerate==1.11.0 bitsandbytes==0.48.1; \
 fi
 

app.py

@@ -1,4 +1,5 @@
 import os
+import spaces
 from pathlib import Path
 
 import gradio as gr

pre-requirements.txt

@@ -0,0 +1,4 @@
+# --- PyTorch (CUDA 12.6) ---
+--extra-index-url https://download.pytorch.org/whl/cu126
+torch>=2.5.1, <=2.8.0
+torchvision>=0.20.1, <=0.24.0

pyproject.toml

@@ -61,8 +61,8 @@ paddle = [
 
 # Extra dependencies for VLM models
 vlm = [
-    "torch==2.6.0", # should use --index-url https://download.pytorch.org/whl/cu126 for cuda support for paddleocr, need to install manually
-    "torchvision==0.21",
+    "torch==2.8.0", # should use --index-url https://download.pytorch.org/whl/cu126 for cuda support for paddleocr, need to install manually
+    "torchvision==0.24.0",
     "transformers==4.57.1",
     "accelerate==1.11.0",
 ]

requirements.txt

@@ -11,24 +11,6 @@ Faker==37.8.0
 python-levenshtein==0.27.1
 rapidfuzz==3.14.1
 
-# --- Machine learning / NLP ---
-scikit-learn==1.7.2
-spacy==3.8.7
-spaczz==0.6.1
-en_core_web_lg @ https://github.com/explosion/spacy-models/releases/download/en_core_web_lg-3.8.0/en_core_web_lg-3.8.0.tar.gz
-transformers==4.57.1
-accelerate==1.11.0
-
-# --- PyTorch (CUDA 12.6) ---
---extra-index-url https://download.pytorch.org/whl/cu126
-torch
-torchvision
-
-# --- PaddleOCR (CUDA 12.6) ---
---extra-index-url https://www.paddlepaddle.org.cn/packages/stable/cu126/
-paddlepaddle-gpu==3.2.1
-paddleocr==3.3.0
-
 # --- PDF / OCR / Redaction tools ---
 pdfminer.six==20250506
 pdf2image==1.17.0
@@ -39,6 +21,9 @@ presidio_analyzer==2.2.360
 presidio_anonymizer==2.2.360
 presidio-image-redactor==0.0.57
 
+# --- Document generation ---
+python-docx==1.2.0
+
 # --- Gradio and apps ---
 gradio==5.49.1
 https://github.com/seanpedrick-case/gradio_image_annotator/releases/download/v0.3.3/gradio_image_annotation-0.3.3-py3-none-any.whl  # Custom annotator version with rotation, zoom, labels, and box IDs
@@ -47,9 +32,20 @@ spaces==0.42.1
 # --- AWS Lambda runtime ---
 awslambdaric==3.1.1
 
-# --- Document generation ---
-python-docx==1.2.0
+# --- Machine learning / NLP ---
+scikit-learn==1.7.2
+spacy==3.8.7
+spaczz==0.6.1
+en_core_web_lg @ https://github.com/explosion/spacy-models/releases/download/en_core_web_lg-3.8.0/en_core_web_lg-3.8.0.tar.gz
 
 # --- Testing ---
 pytest>=7.0.0
-pytest-cov>=4.0.0
+pytest-cov>=4.0.0
+
+transformers==4.57.1
+accelerate==1.11.0
+
+# --- PaddleOCR (CUDA 12.6) ---
+--extra-index-url https://www.paddlepaddle.org.cn/packages/stable/cu126/
+paddlepaddle-gpu>=3.0.0, <=3.3.0
+paddleocr<=3.3.0