Logging to UI

app.py

@@ -1,4 +1,5 @@
 import io
+import logging
 import os
 import uuid
 
@@ -13,6 +14,11 @@ from transformers.image_utils import load_image
 
 DEVICE = "cuda" if torch.cuda.is_available() else "cpu"
 
+# Capture logs
+log_stream = io.StringIO()
+logging.basicConfig(stream=log_stream, level=logging.INFO)
+
+
 if "session_id" not in st.session_state:
     st.session_state["session_id"] = str(uuid.uuid4())  # Generate unique session ID
 
@@ -58,6 +64,8 @@ with st.sidebar:
     "[Source Code](https://huggingface.co/spaces/deepakkarkala/multimodal-rag/tree/main)"
 
 st.title("📝 Image Q&A with VLM")
+st.text_area("Logs:", log_stream.getvalue(), height=200)
+
 uploaded_pdf = st.file_uploader("Upload PDF file", type=("pdf"))
 query = st.text_input(
     "Ask something about the image",
@@ -77,13 +85,17 @@ if uploaded_pdf and "is_index_complete" not in st.session_state:
     model_embedding.index(
         input_path=images_folder, index_name=index_name, store_collection_with_index=False, overwrite=True
     )
+    logging.info(f"{len(images)} number of images extracted from PDF and indexed")
     st.session_state["is_index_complete"] = True
 
 
 
 if uploaded_pdf and query:
     docs_retrieved = model_embedding.search(query, k=1)
-    image_similar_to_query = images[docs_retrieved[0]["doc_id"]]
+    logging.info(f"{len(docs_retrieved)} number of images retrieved as relevant to query")
+    image_id = docs_retrieved[0]["doc_id"]
+    logging.info(f"Image id:{image_id} retrieved" )
+    image_similar_to_query = images[image_id]
 
     # Create input messages
     system_prompt = "You are an AI assistant. Your task is reply to user questions based on the provided image context."

requirements.txt

@@ -1,11 +1,6 @@
 streamlit>=1.28
-langchain>=0.0.217
-openai>=1.2
-duckduckgo-search
-anthropic>=0.3.0
 trubrics>=1.4.3
 streamlit-feedback
-langchain-community
 torch
 transformers
 accelerate>=0.26.0