Merge branch 'main' into dependabot/pip/ai/gen-ai-agents/travel_agent/pillow-11.3.0

AlexanderHodicke · web-flow · commit 5017d6461a30 · 2025-07-02T10:38:01.000+02:00
diff --git a/ai/generative-ai-service/smart-invoice-extraction/SmartInvoiceExtraction.py b/ai/generative-ai-service/smart-invoice-extraction/SmartInvoiceExtraction.py
@@ -1,16 +1,57 @@
-import pandas as pd
+"""
+Smart Invoice Extraction
+
+A Streamlit-based invoice data extraction tool leveraging Oracle Cloud Infrastructure (OCI)
+Generative AI models for multimodal (text + image) processing.
+
+This module provides:
+
+- Helper functions:
+  - save_images(images, output_format="JPEG"):
+      Convert PIL Image objects to in-memory byte streams for downstream processing.
+  - encode_image(image_path):
+      Read an image file from disk and return its Base64-encoded string.
+  - save_to_csv(data, file_name="extracted_data.csv"):
+      Persist a list of dicts to a CSV file.
+
+- extractor(image_list):
+    Uses the image of a PDF invoice to identify and extract key header fields, by:
+      • Initializing an LLM (meta.llama-3.2-90b-vision-instruct)
+      • Encoding the image as Base64
+      • Sending a system + human message prompt to extract invoice headers in list format.
+
+- invoiceAnalysisPlus():
+    The main Streamlit application which:
+      • Renders a UI for uploading PDF invoices
+      • Converts PDFs to JPEG images and prepares byte streams
+      • Invokes extractor() to propose candidate fields
+      • Constructs and sends prompts to two OCI LLMs:
+          – A vision model for image-based extraction
+          – A text-only model (cohere.command-r-plus-08-2024) for dynamic prompt generation
+      • Displays results in a DataFrame and saves them to CSV
+
+Usage:
+    Run the module as a standalone script:
+        streamlit run invoice_analysis.py
+
+Author: Ali Ottoman
+"""
+
+import io
 import json
-from langchain.chains.llm import LLMChain
-from langchain_core.prompts import PromptTemplate
+import base64
+import pandas as pd
 import streamlit as st
 from langchain_community.chat_models.oci_generative_ai import ChatOCIGenAI
 from langchain_core.messages import HumanMessage, SystemMessage
-import base64
 from pdf2image import convert_from_bytes
-import io
+
 
 # Helper function to convert a list of images into byte arrays for further processing
 def save_images(images, output_format="JPEG"):
+    """
+        Convert PIL Image objects to in-memory byte streams for downstream processing.
+    """
     image_list = []
     for image in images:
         img_byte_arr = io.BytesIO()
@@ -21,17 +62,26 @@ def save_images(images, output_format="JPEG"):
 
 # Helper function to encode an image to base64 for sending to LLM
 def encode_image(image_path):
+    """
+        Read an image file from disk and return its Base64-encoded string.
+    """
     with open(image_path, "rb") as image_file:
         return base64.b64encode(image_file.read()).decode("utf-8")
 
 # Save extracted data to a CSV file and show success message in Streamlit
 def save_to_csv(data, file_name="extracted_data.csv"):
+    """
+        Persist a list of dicts to a CSV file.
+    """
     df = pd.DataFrame(data)
     df.to_csv(file_name, index=False)
     st.success(f"Data saved to {file_name}")
 
 # Extract key headers from the first image of a PDF invoice
 def extractor(image_list):
+    """
+        Uses the image of a PDF invoice to identify and extract key header fields
+    """
     # Replace this with your own compartment ID
     compID = "<YOUR_COMPARTMENT_OCID_HERE>"
 
@@ -68,155 +118,156 @@ def extractor(image_list):
         return eval(ai_response.content)
 
 # Main Streamlit app function
-def invoiceAnalysisPlus():
+def invoice_analysis_plus():
+    """
+        The main Streamlit application
+    """
     st.title("Invoice Data Extraction")
-    
+
     with st.sidebar:
         st.title("Parameters")
-        # User prompt input
+        # Replace with your own compartment ID
+        compID = "<YOUR_COMPARTMENT_OCID_HERE>"
         user_prompt = st.text_input("Input the elements you are looking to extract here")
         st.caption("Our AI assistant has extracted the following key elements from the invoice. Please select the elements you wish to extract.")
 
-        
     uploaded_file = st.file_uploader("Upload your invoices here:", type=["pdf"])
-    
+
     if uploaded_file is not None:
         with st.spinner("Processing..."):
+            # Convert PDF to image list
             if uploaded_file.type == "application/pdf":
                 images = convert_from_bytes(uploaded_file.read(), fmt="jpeg")
             else:
                 images = [convert_from_bytes(uploaded_file.read(), fmt="jpeg")[0]]
-            
-            image_list = save_images(images)  # Convert to byte arrays
-        
+
+            # Save as byte streams
+            image_list = save_images(images)
+
+        # Load both image-based and text-based LLMs
         llm = ChatOCIGenAI(
-            model_id="meta.llama-3.2-90b-vision-instruct",
-            compartment_id="", #TO-DO: Add your compartment ID here
+            model_id="meta.llama-3.2-90b-vision-instruct",  # Replace with your model ID
+            compartment_id=compID,
             model_kwargs={"max_tokens": 2000, "temperature": 0}
         )
         llm_for_prompts = ChatOCIGenAI(
-            model_id="cohere.command-r-plus-08-2024",
-            compartment_id="",#TO-DO: Add your compartment ID here
+            model_id="cohere.command-r-plus-08-2024",  # Replace with your model ID
+            compartment_id=compID,
             model_kwargs={"max_tokens": 2000, "temperature": 0}
         )
-        
-        # Options for data types
-        data_types = [ "Text", "Number", "Percentage", "Date"]
-        
-        # Lists to store names and their types
+
+        # Select box UI for user to pick elements and their data types
+        data_types = ["Text", "Number", "Percentage", "Date"]
         elements = []
+
         if "availables" not in st.session_state:
             st.session_state.availables = extractor(image_list)
-        for i in range(3):  # Adjust 'n' for the maximum number of selections
-            col1, col2 = st.columns([2, 1])  # Adjust width ratio if needed
-            
-            with col1:                
-                # Preserve user selection across reruns
+
+        for i in range(3):  # Max 3 fields
+            col1, col2 = st.columns([2, 1])
+            with col1:
                 name = st.selectbox(f"Select an element {i+1}", st.session_state.availables, key=f"name_{i}", index=i)
             with col2:
                 data_type = st.selectbox(f"Type {i+1}", data_types, key=f"type_{i}")
                 elements.append((name, data_type))
 
-        if elements is not None:
+        # Generate appropriate prompt based on selected or input fields
+        if elements:
             system_message_cohere = SystemMessage(
-            content=f"""
-            Based on the following set of elements {elements}, with their respective types ({elements[0][1]}, {elements[1][1]}, {elements[2][1]}), Extract the following details and provide the response only in valid JSON format (no extra explanation or text):
-            - {elements[0][0]}
-            - {elements[1][0]}
-            - {elements[2][0]}
-            Ensure the extracted data is formatted correctly as JSON and include nothing else at all in the response, not even a greeting or closing.
-            For example:
-            {{
-                {elements[0][0]}: "296969",
-                {elements[1][0]}: "296969",
-                {elements[2][0]}: "296969",
-            }}
-            """)
+                content=f"""
+                    Based on the following set of elements {elements}, with their respective types ({elements[0][1]}, {elements[1][1]}, {elements[2][1]}), Extract the following details and provide the response only in valid JSON format (no extra explanation or text):
+                    - {elements[0][0]}
+                    - {elements[1][0]}
+                    - {elements[2][0]}
+                    Ensure the extracted data is formatted correctly as JSON and include nothing else at all in the response, not even a greeting or closing.
+                    For example:
+                    {{
+                        {elements[0][0]}: "296969",
+                        {elements[1][0]}: "296969",
+                        {elements[2][0]}: "296969",
+                    }}
+                """)
             ai_response_cohere = system_message_cohere
         else:
-            # Cohere section for generating the prompt
             system_message_cohere = SystemMessage(
-            content=f"""
-            Based on the following system prompt, create a new prompt accordingly based on the elements specified in the user prompt here ({user_prompt}). 
-
-            This is the system prompt template:
-            "
-            Extract the following details and provide the response only in valid JSON format (no extra explanation or text):
-            - **Debit / Credit Note No.**
-            - **Policy Period** 
-            - **Insured** 
-            - **Vessel Name** 
-            - **Details** 
-            - **Currency** 
-            - **Gross Premium 100%**
-            - **OIMSL Share** 
-            - **Total Deductions**
-            - **Net Premium** 
-            - **Premium Schedule**
-            - **Installment Amount**
-
-            Ensure the extracted data is formatted correctly as JSON and include nothing else at all in the response, not even a greeting or closing.
-
-            For example:
-            
-                "Debit / Credit Note No.": "296969",
-                "Policy Period": "Feb 20, 2024 to Jul 15, 2025",
-                "Insured": "Stealth Maritime Corp. S.A.",
-                "Vessel Name": "SUPRA DUKE - HULL & MACHINERY", (Make sure this is the entire vessel name only)
-                "Details": "SUPRA DUKE - Original Premium",
-                "Currency": "USD",
-                "Gross Premium 100%": 56973.63,
-                "OIMSL Share": 4557.89,
-                "Total Deductions": 979.92,
-                "Net Premium": 3577.97,
-                "Premium Schedule": ["Apr 20, 2024", "Jun 14, 2024", "Sep 13, 2024", "Dec 14, 2024", "Mar 16, 2025", "Jun 14, 2025"],
-                "Installment Amount": [372.87, 641.02, 641.02, 641.02, 641.02, 641.02]
-            
-            )" ensure your response is a system prompt format with an example of what the ouput should look like. Also ensure to mention in your gernerated prompt that no other content whatsover should appear except the JSON
+                content = f"""
+                    Based on the following system prompt, create a new prompt accordingly based on the elements specified in the user prompt here ({user_prompt}). 
+
+                    This is the system prompt template:
+                    "
+                    Extract the following details and provide the response only in valid JSON format (no extra explanation or text):
+                    - **Debit / Credit Note No.**
+                    - **Policy Period** 
+                    - **Insured** 
+                    - **Vessel Name** 
+                    - **Details** 
+                    - **Currency** 
+                    - **Gross Premium 100%**
+                    - **OIMSL Share** 
+                    - **Total Deductions**
+                    - **Net Premium** 
+                    - **Premium Schedule**
+                    - **Installment Amount**
+
+                    Ensure the extracted data is formatted correctly as JSON and include nothing else at all in the response, not even a greeting or closing.
+
+                    For example:
+                    
+                        "Debit / Credit Note No.": "296969",
+                        "Policy Period": "Feb 20, 2024 to Jul 15, 2025",
+                        "Insured": "Stealth Maritime Corp. S.A.",
+                        "Vessel Name": "SUPRA DUKE - HULL & MACHINERY", (Make sure this is the entire vessel name only)
+                        "Details": "SUPRA DUKE - Original Premium",
+                        "Currency": "USD",
+                        "Gross Premium 100%": 56973.63,
+                        "OIMSL Share": 4557.89,
+                        "Total Deductions": 979.92,
+                        "Net Premium": 3577.97,
+                        "Premium Schedule": ["Apr 20, 2024", "Jun 14, 2024", "Sep 13, 2024", "Dec 14, 2024", "Mar 16, 2025", "Jun 14, 2025"],
+                        "Installment Amount": [372.87, 641.02, 641.02, 641.02, 641.02, 641.02]
+                    
+                    )" ensure your response is a system prompt format with an example of what the ouput should look like. Also ensure to mention in your gernerated prompt that no other content whatsover should appear except the JSON
             """)
             ai_response_cohere = llm_for_prompts.invoke(input=[system_message_cohere])
-        print(ai_response_cohere)
 
+        # Extracted data list
         extracted_data = []
-        
+
         with st.spinner("Analyzing invoice..."):
             for idx, img_byte_arr in enumerate(image_list):
                 try:
-                    # Convert the image to base64 directly from memory
                     encoded_frame = base64.b64encode(img_byte_arr.getvalue()).decode("utf-8")
-                    if elements is not None:
+
+                    if elements:
                         system_message = ai_response_cohere
                     else:
-                        system_message = SystemMessage(
-                            content=ai_response_cohere.content)
+                        system_message = SystemMessage(content=ai_response_cohere.content)
+
                     human_message = HumanMessage(
                         content=[
                             {"type": "text", "text": "This is my invoice"},
                             {"type": "image_url", "image_url": {"url": f"data:image/jpeg;base64,{encoded_frame}"}},
                         ]
                     )
+
                     ai_response = llm.invoke(input=[human_message, system_message])
-                    print(ai_response.content)
-                    index = ai_response.content.find('{')
-                    index2 = ai_response.content.find('}')
-                    x = ai_response.content[index:]
-                    x2 = x[:index2+1]
-                    print(x2)
-                    response_dict = json.loads(x2)
-                    
-                    # Add metadata for tracking
-                    response_dict["File Name"] = uploaded_file.name
-                    response_dict["Page Number"] = idx + 1  
+                    json_start = ai_response.content.find('{')
+                    json_end = ai_response.content.find('}', json_start)
+                    json_data = ai_response.content[json_start:json_end + 1]
 
+                    response_dict = json.loads(json_data)
+                    response_dict["File Name"] = uploaded_file.name
+                    response_dict["Page Number"] = idx + 1
                     extracted_data.append(response_dict)
 
                 except Exception as e:
                     st.error(f"Error processing page {idx+1}: {str(e)}")
-                    
+
+        # Display and save results
         if extracted_data:
             save_to_csv(extracted_data)
             st.dataframe(pd.DataFrame(extracted_data))
 
-# Run the chatbot function
+# Run the app
 if __name__ == "__main__":
-    invoiceAnalysisPlus()
+    invoice_analysis_plus()
diff --git a/cloud-infrastructure/ai-infra-gpu/ai-infrastructure/nemo-megatron-training-oke/README.md b/cloud-infrastructure/ai-infra-gpu/ai-infrastructure/nemo-megatron-training-oke/README.md
@@ -8,7 +8,7 @@ on the Oracle Container Engine for Kubernetes (OKE) using
 Reference results from NVIDIA to train Llama 3 can be found on the
 [NGC Catalog](https://catalog.ngc.nvidia.com/orgs/nvidia/teams/dgxc-benchmarking/resources/llama3-dgxc-benchmarking).
 
-Reviewed: 18.03.2025
+Reviewed: 01.07.2025
 
 # When to use this asset?
 
@@ -31,7 +31,14 @@ This guide is loosely based on the
    [to the instructions](https://github.com/oracle-quickstart/oci-hpc-oke/tree/main#instructions-for-deploying-an-oke-cluster-with-gpus-and-rdma-connectivity),
    importing one of the images and creating a GPU partition with BM.GPU.H100.8 nodes.
 
-   The configuration here assumes a minimum of 16 BM.GPU.H100.8 nodes.
+   The configuration here assumes a minimum of 1 BM.GPU.H100.8 node for
+   training with 8B parameters, and a minimum of 8 BM.GPU.H100.8 nodes for 70B
+   parameters.
+   
+   If another shape is used, the NCCL and MPI parameters in the Kubernetes
+   [configuration map](./files/training/templates/mpi.yaml) should be adapted
+   using the same parameter values as the
+   [performance testing scripts](https://github.com/oracle-quickstart/oci-hpc-oke/tree/main/manifests/nccl-tests).
 
    - Ensure that the follwing setting is selected under the "OKE Cluster" section:
 
diff --git a/cloud-infrastructure/compute-including-hpc/compute-software/README.md b/cloud-infrastructure/compute-including-hpc/compute-software/README.md
@@ -20,6 +20,7 @@ This page contains information and useful links regarding Compute services that
 - [Youtube Video: Using a Network Bridge with KVM VMs on Oracle Linux](https://www.youtube.com/watch?v=CXBTBxFoSKI&t=120s)
 - [Youtube Video: Manage OCI instances directly from VirtualBox 7](https://www.youtube.com/watch?v=uFEN4Di-WDE)
 
+- [Olygo Github: OCI_Compute_Function_Reserved_Pip_Allocator](https://github.com/Olygo/OCI-FN_reserved_pip_allocator)
 - [Olygo Github: OCI Compute Capacity Report using CloudShell](https://github.com/Olygo/OCI_ComputeCapacityReport)
 - [Olygo Github: Learn how to troubleshoot Linux and Windows instances using OCI Console Connection](https://github.com/Olygo/OCI_Console-Connections)
 - [Olygo Github: OCI function that forces IMDSv2 on compute instances](https://github.com/Olygo/OCI-FN_IMDS-Watcher)
