Upload 2 files

Dockerfile

@@ -1,5 +1,5 @@
-# Use an official Python runtime as the base image
-FROM python:3.11-slim
+# Use an official Python runtime matching Hugging Face's environment
+FROM python:3.10-slim
 
 # Set working directory
 WORKDIR /app
@@ -13,17 +13,15 @@ RUN apt-get update && apt-get install -y \
 # Install Ollama
 RUN curl -fsSL https://ollama.com/install.sh | sh
 
-########################################################
-# Pull llama3 model during build to avoid runtime delays
+# Pre-pull llama3 during build to avoid runtime delays
 RUN ollama serve & \
     until curl -s http://localhost:11434 > /dev/null; do \
         echo 'Waiting for Ollama...'; sleep 1; \
     done && \
     ollama pull llama3 && \
+    echo "Model pulled successfully" || echo "Model pull failed" && \
     ollama list > /app/models.txt && \
     cat /app/models.txt
-########################################################
-
 
 # Copy requirements file first (optimization for caching)
 COPY requirements.txt .
@@ -31,7 +29,7 @@ COPY requirements.txt .
 # Install Python dependencies
 RUN pip install --no-cache-dir -r requirements.txt
 
-# Copy only necessary application files (exclude .env)
+# Copy only the app file
 COPY app.py .
 
 # Expose the port Hugging Face Spaces expects
@@ -41,5 +39,5 @@ EXPOSE 7860
 ENV OLLAMA_HOST=0.0.0.0
 ENV OLLAMA_PORT=11434
 
-# Start Ollama and run Streamlit
-CMD bash -c "ollama serve & until curl -s http://localhost:11434 > /dev/null; do echo 'Waiting for Ollama...'; sleep 1; done && streamlit run app.py --server.port 7860 --server.address 0.0.0.0"
+# Start Ollama and Streamlit with a more robust wait
+CMD bash -c "ollama serve & sleep 5 && until curl -s http://localhost:11434 > /dev/null; do echo 'Waiting for Ollama...'; sleep 1; done && streamlit run app.py --server.port 7860 --server.address 0.0.0.0"

app.py

@@ -6,8 +6,13 @@ from atomic_agents.lib.components.system_prompt_generator import SystemPromptGen
 from atomic_agents.agents.base_agent import BaseAgent, BaseAgentConfig, BaseAgentInputSchema, BaseAgentOutputSchema
 from dotenv import load_dotenv
 import asyncio
+import logging
 
-# Load environment variables (optional if using Hugging Face Secrets)
+# Set up logging
+logging.basicConfig(level=logging.INFO)
+logger = logging.getLogger(__name__)
+
+# Load environment variables (optional for Hugging Face Secrets)
 load_dotenv()
 
 # Initialize Streamlit app
@@ -20,18 +25,24 @@ def setup_client(provider):
         from openai import AsyncOpenAI
         api_key = os.getenv("OPENAI_API_KEY")
         if not api_key:
-            st.warning("OpenAI provider unavailable: OPENAI_API_KEY not set. Falling back to Ollama.")
-            return setup_client("ollama")  # Fallback to Ollama
+            st.warning("OpenAI unavailable: OPENAI_API_KEY not set. Using Ollama.")
+            return setup_client("ollama")
         client = instructor.from_openai(AsyncOpenAI(api_key=api_key))
         model = "gpt-4o-mini"
         display_model = "OpenAI (gpt-4o-mini)"
     elif provider == "ollama":
         from openai import AsyncOpenAI as OllamaClient
-        client = instructor.from_openai(
-            OllamaClient(base_url="http://localhost:11434/v1", api_key="ollama"), mode=instructor.Mode.JSON
-        )
-        model = "llama3"
-        display_model = "Ollama (llama3)"
+        try:
+            client = instructor.from_openai(
+                OllamaClient(base_url="http://localhost:11434/v1", api_key="ollama"), mode=instructor.Mode.JSON
+            )
+            model = "llama3"
+            display_model = "Ollama (llama3)"
+            logger.info("Ollama client initialized successfully")
+        except Exception as e:
+            logger.error(f"Failed to initialize Ollama client: {e}")
+            st.error(f"Ollama connection failed: {e}")
+            return None, None, None
     else:
         st.error(f"Unsupported provider: {provider}")
         return None, None, None
@@ -50,7 +61,7 @@ system_prompt_generator = SystemPromptGenerator(
 )
 
 # Provider selection
-providers_list = ["ollama", "openai"]  # Prioritize Ollama since it’s guaranteed to work
+providers_list = ["ollama", "openai"]
 selected_provider = st.selectbox("Choose a provider:", providers_list, key="provider_select")
 
 # Set up client and agent based on the selected provider
@@ -73,7 +84,7 @@ if "agent" not in st.session_state or st.session_state.get("current_model") != m
         memory=st.session_state.memory,
         system_role="developer",
     ))
-    st.session_state.current_model = model  # Track the current model to detect changes
+    st.session_state.current_model = model
 
 # Display the selected model
 st.markdown(f"**Selected Model:** {st.session_state.display_model}")
@@ -107,11 +118,15 @@ if user_input:
         response_container = st.empty()
         async def stream_response():
             current_response = ""
-            async for partial_response in st.session_state.agent.run_async(input_schema):
-                if hasattr(partial_response, "chat_message") and partial_response.chat_message:
-                    if partial_response.chat_message != current_response:
-                        current_response = partial_response.chat_message
-                        response_container.markdown(current_response)
+            try:
+                async for partial_response in st.session_state.agent.run_async(input_schema):
+                    if hasattr(partial_response, "chat_message") and partial_response.chat_message:
+                        if partial_response.chat_message != current_response:
+                           current_response = partial_response.chat_message
+                           response_container.markdown(current_response)
+            except Exception as e:
+                logger.error(f"Error streaming response: {e}")
+                response_container.error(f"Error: {e}")
 
             # After streaming completes, add the final response to conversation and memory
             st.session_state.conversation.append(("assistant", current_response))