Original upload of app.py, requirements.txt, and tesla_db (vector database) folder from a local computer folder (not directly from Google Drive folder)

.gitattributes

@@ -33,3 +33,4 @@ saved_model/**/* filter=lfs diff=lfs merge=lfs -text
 *.zip filter=lfs diff=lfs merge=lfs -text
 *.zst filter=lfs diff=lfs merge=lfs -text
 *tfevents* filter=lfs diff=lfs merge=lfs -text
+tesla_db/chroma.sqlite3 filter=lfs diff=lfs merge=lfs -text

app.py

@@ -0,0 +1,181 @@
+import os
+import uuid
+import json
+from pathlib import Path
+
+# ATTENTION: some versions in the requirements.txt file are more current than the ones used in the notebook
+
+# GUI components
+import gradio as gr
+
+# Accessing external environment with endpoint and secret
+# Using openai as a dummy container to connect to an endpoint and send HTTP requests (note: one could also use the python "request" package functions)
+# Accessing OpenAI Model hosting platform
+from openai import OpenAI, OpenAIError
+
+# Embedding operations & Vector DB creation
+from langchain_community.embeddings.sentence_transformer import SentenceTransformerEmbeddings
+from langchain_community.vectorstores import Chroma
+
+# HuggingFace platform
+from huggingface_hub import CommitScheduler
+
+# Set working directory (HuggingFace space)
+hf_space_dir = os.getcwd()
+
+# Anyscale model hosting platform NOT USED in this notebook.  Open AI model hosting platform used instead.
+# client = OpenAI(
+#     base_url="https://api.endpoints.anyscale.com/v1",
+#     api_key=os.environ['ANYSCALE_API_KEY']
+# )
+
+#----------------------------------------------------------------------
+# OpenAI API key stored as a "secret" HuggingFace Space
+# OPENAI_API_KEY 
+
+#import os
+# Set your API key from an environment variable: get("OPENAI_API_KEY")
+# app.py should have access to the OpenAI key stored as a HuggingFace secret in the HuggingFace space where the app.py will be uploaded, by accessing it through the environment variables
+# openai_api_key = os.getenv("OpenAI_API_key_GL_Adv_Python_Project")  
+# or:
+# openai_api_key = os.environ.get("OpenAI_API_key_GL_Adv_Python_Project")
+# or:
+openai_api_key = os.environ("OpenAI_API_key_GL_Adv_Python_Project")    # environ is a dictionary with all the environment variables, so to access the value, provide the key
+
+# Set Up Your API Key: You'll need an API key from OpenAI. You can obtain one by signing up on the OpenAI website and navigating to your API keys in the dashboard.
+# This approach ensures that your API key remains secure and is not exposed in your code.
+
+# OpenAI models:
+
+# model_name = "gpt-4o"           # Cost:  Input: $5 / 1M tokens     ;  Output: $15 / 1M tokens
+                                # GPT-4o is OpenAI's most advanced multimodal model that’s faster and cheaper than GPT-4 Turbo with stronger vision capabilities.
+                                # The model has 128K context and an October 2023 knowledge cutoff.
+
+model_name = "gpt-4o-mini"      # Cost:  Input: $0.15 / 1M tokens  ;  Output: $0.60 / 1M tokens
+                                # GPT-4o mini is our most cost-efficient small model that’s smarter and cheaper than GPT-3.5 Turbo, and has vision capabilities.
+                                # The model has 128K context and an October 2023 knowledge cutoff.
+
+# Create an OpenAI Client: setting up the client with new version of OpenAI Python library - version OpenAI 1.0.0 and above
+
+client = OpenAI(
+    # This is the default and can be omitted
+    # api_key=os.environ.get("OPENAI_API_KEY"),
+    api_key=openai_api_key,
+)
+#---------------------------------------------------------------------
+
+embedding_model = SentenceTransformerEmbeddings(model_name='thenlper/gte-small')
+
+tesla_10k_collection = 'tesla-10k-2019-to-2023'
+
+# vector database constructor Chroma()
+vectorstore_persisted = Chroma(
+    collection_name=tesla_10k_collection,
+    persist_directory='./tesla_db',
+    embedding_function=embedding_model
+)
+
+retriever = vectorstore_persisted.as_retriever(
+    search_type='similarity',
+    search_kwargs={'k': 5}
+)
+
+# Prepare the logging functionality
+
+log_file = Path("logs/") / f"data_{uuid.uuid4()}.json"
+log_folder = log_file.parent
+
+scheduler = CommitScheduler(
+    repo_id="document-qna-chroma-openai-logs",      # name of the log folder containing json elements -->  HuggingFace dataset       # OLD name: "document-qna-chroma-anyscale-logs",
+    repo_type="dataset",
+    folder_path=log_folder,
+    path_in_repo="data",
+    every=2                                         # execute every two API calls
+
+qna_system_message = """
+You are an assistant to a financial services firm who answers user queries on annual reports.
+Users will ask questions delimited by triple backticks, that is, ```.
+User input will have the context required by you to answer user questions.
+This context will begin with the token: ###Context.
+The context contains references to specific portions of a document relevant to the user query.
+Please answer only using the context provided in the input. However, do not mention anything about the context in your answer. 
+If the answer is not found in the context, respond "I don't know".
+"""
+
+qna_user_message_template = """
+###Context
+Here are some documents that are relevant to the question.
+{context}
+```
+{question}
+```
+"""
+
+# Define the predict function that runs when 'Submit' is clicked or when a API request is made
+def predict(user_input):
+
+    relevant_document_chunks = retriever.invoke(user_input)
+    # relevant_document_chunks = retriever.get_relevant_documents(query = user_input)
+    context_list = [doc_chunk.page_content for doc_chunk in relevant_document_chunks]
+    context_for_query = ". ".join(context_list)
+    
+    prompt = [
+        {'role':'system', 'content': qna_system_message},
+        {'role':'user', 'content': qna_user_message_template.format(
+            context=context_for_query,
+            question=user_input
+            )
+        }
+    ]
+
+    try:
+        response = client.chat.completions.create(
+            model=model_name,                # previous model used: 'mlabonne/NeuralHermes-2.5-Mistral-7B',
+            messages=prompt,
+            temperature=0,                   # Temperature > 0 to encourage creative answer... Temperature = 0.7: A common setting that provides a balance between creativity and coherence.
+            # max_tokens=200                 # Limit the number of tokens in the response 
+        )
+
+        prediction = response.choices[0].message.content.strip()   # Access response attributes directly
+
+    except Exception as e:
+        prediction = e
+
+    # While the prediction is made, log both the inputs and outputs to a local log file (i.e., HuggingFace dataset)
+    # While writing to the log file, ensure that the commit scheduler is locked to avoid parallel
+    # access (i.e., put a lock on the state of the log_file in case user are entering queries while the log operation is in progress.)
+    # Note: the log_file is a json file.
+    
+    with scheduler.lock:
+        with log_file.open("a") as f:
+            # json.dumps turns the dictionary into a json string containing 'user_input', 'context_for_query', and 'prediction'
+            f.write(json.dumps(
+                {
+                    'user_input': user_input,
+                    'retrieved_context': context_for_query,
+                    'model_response': prediction
+                }
+            ))
+            f.write("\n")     # write a new line to prepare for the next observation to be logged
+    
+    return prediction
+
+
+textbox = gr.Textbox(placeholder="Enter your query here", lines=6)
+
+# Create the interface
+demo = gr.Interface(
+    inputs=textbox, fn=predict, outputs="text",
+    title="Ask Me Anything (AMA) on Tesla 10-K statements",
+    description="This web API presents an interface to ask questions on contents of the Tesla 10-K reports for the period 2019 - 2023.",
+    article="Note that questions that are not relevant to the Tesla 10-K report will not be answered.",
+    examples=[["What was the total revenue of the company in 2022?", "$ 81.46 Billion"],
+            ["Summarize the Management Discussion and Analysis section of the 2021 report in 50 words.", ""],
+            ["What was the company's debt level in 2020?", ""],
+            ["Identify 5 key risks identified in the 2019 10k report? Respond with bullet point summaries.", ""]
+    ],
+    concurrency_limit=16
+)
+
+demo.queue()
+demo.launch()

requirements.txt

@@ -0,0 +1,5 @@
+openai==1.23.2
+chromadb==0.4.22
+langchain==0.1.9
+langchain-community==0.0.32
+sentence-transformers==2.3.1

tesla_db/3b5f5183-1f48-4873-ae05-297f75d85ed2/data_level0.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:f9ff9f9ee0095bf3d2a87e98c76eb56a3f34fb13c041cd7fcfc43dc6b85cc4d3
+size 12708000

tesla_db/3b5f5183-1f48-4873-ae05-297f75d85ed2/header.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:e0e8e145fcc03c24c5125dd89060c5951b0abcf71bb72771d6bab0efefa08aa0
+size 100

tesla_db/3b5f5183-1f48-4873-ae05-297f75d85ed2/index_metadata.pickle

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:373446c85bb79dec763010e5b33a241e2b9cfacc8c1a119dca0fdc45adc0d1e7
+size 172004

tesla_db/3b5f5183-1f48-4873-ae05-297f75d85ed2/length.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:69f0dec43639a3ed312ef382157c5d9611ae2ee5a872375b6d5fd8d00622c750
+size 12000

tesla_db/3b5f5183-1f48-4873-ae05-297f75d85ed2/link_lists.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:e2d324b759e78b46a63ae38998cec57587475ed8ea1c2a2e688b87a957072c21
+size 25736

tesla_db/chroma.sqlite3

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:06ed87d12e6ee1b0b1f1f81fdfa4576e32f095c86fb7a2a66d0200a7640da76a
+size 46223360