.gitignore

+*env/
+.env
+
+*.egg-*
+*.pyc
+*.txt
+config.txt
+
+run.sh
+build.sh
+run_my_app.sh
+run_streamlit_app.sh
\ No newline at end of file

.gitlab-ci.yml

 variables:
   DOCKER_TLS_CERTDIR: ""
+  TIMEOUT: 3600
 docker-build-master:
   # Official docker image.
   image: docker:latest
@@ -10,5 +11,5 @@ docker-build-master:
     - docker login -u "$CI_REGISTRY_USER" -p "$CI_REGISTRY_PASSWORD" $CI_REGISTRY
   script:
     - cp $ENV_FILE .env
-    - docker build --pull -t "$CI_REGISTRY_IMAGE":prod .
-    - docker push "$CI_REGISTRY_IMAGE":prod
+    - docker build --pull -t "$CI_REGISTRY_IMAGE":prodcc4 .
+    - docker push "$CI_REGISTRY_IMAGE":prodcc4

Dockerfile

-FROM jupyter/scipy-notebook:hub-1.5.0
+FROM quay.io/jupyter/datascience-notebook:hub-5.2.0
 
-# Install from APT repository
 USER root
-RUN apt-get update -y
-RUN apt-get install -y git libpq-dev gcc
 
-# Set working directory
-WORKDIR /usr/src/app
+COPY requirements.txt .
+RUN pip install --no-cache-dir -r requirements.txt && rm requirements.txt
 
-# Install basics
-USER jovyan
-RUN pip3 install --upgrade pip
-COPY .env .env
-
-# Install python packages
-COPY requirements.txt requirements.txt
-RUN pip install --no-cache-dir -r requirements.txt
-ENV IPYTHONDIR /usr/src/app/ipython/
-WORKDIR /usr/src/app/ipython/profile_default/startup/
-COPY python_startup/ ./
-WORKDIR /home/
+# install some utilities for GPT
+COPY llm_utils /llm_utils/
+RUN pip install /llm_utils/
+ENV CONFIG_PATH=/home/jovyan/config.txt
 
+USER jovyan
\ No newline at end of file

README.md

-# bücker
-
+# A Jupyterlab for LLM

llm_utils/setup.py

+import setuptools
+
+setuptools.setup(
+    author="Julian Rasch",
+    author_email="julian.rasch@fh-muenster.de",
+    description="Helper modules to work with LLMs.",
+    name="llm_utils",
+    package_dir={"": "src"},
+    packages=setuptools.find_packages(where="src"),
+    install_requires=[
+        "openai", 
+        "python-dotenv"
+    ]
+)

llm_utils/src/llm_utils/client.py

+import os
+from openai import AzureOpenAI
+from dotenv import load_dotenv
+
+from enum import Enum
+
+
+class OpenAIModels(Enum):
+    GPT_3 = "gpt3"
+    GPT_4 = "gpt4"
+    GPT_4o = "gpt-4o"
+    EMBED = "embed"
+
+    @classmethod
+    def get_all_values(cls):
+        return [member.value for member in cls]
+
+
+def get_openai_client(
+    model: str,
+    config_path: str
+    ) -> AzureOpenAI:
+    if not model in OpenAIModels.get_all_values():
+        raise ValueError(f"<model> needs to be one of {OpenAIModels.get_all_values()}.")
+    
+    load_dotenv(
+        dotenv_path=config_path,
+        override=True
+    )
+
+    AZURE_OPENAI_API_KEY = os.environ.get("AZURE_OPENAI_API_KEY")
+    AZURE_OPENAI_ENDPOINT = os.environ.get("AZURE_OPENAI_ENDPOINT")
+    OPENAI_API_VERSION = os.environ.get("OPENAI_API_VERSION")
+    
+    if any(p is None for p in (AZURE_OPENAI_API_KEY, AZURE_OPENAI_API_KEY, OPENAI_API_VERSION)):
+        raise ValueError(
+            f"""None of the following parameters can be None: 
+            AZURE_OPENAI_API_KEY: {AZURE_OPENAI_API_KEY},
+            AZURE_OPENAI_API_KEY: {AZURE_OPENAI_API_KEY},
+            OPENAI_API_VERSION: {OPENAI_API_VERSION}
+            """
+        )
+    
+    client = AzureOpenAI(
+        api_key=AZURE_OPENAI_API_KEY,
+        azure_endpoint=AZURE_OPENAI_ENDPOINT,
+        # api_version=OPENAI_API_VERSION, 
+        azure_deployment=model
+    )
+    return client
+
+
+class ChatGPT:
+    def __init__(self, client: AzureOpenAI, model: str):
+        self.model = model
+        self.client = client
+        self.messages = []
+
+    def chat_with_gpt(self, user_input: str):
+        self.messages.append({
+            "role": "user",
+            "content": user_input
+        })
+        response = self._generate_response(self.messages)
+        return response
+
+    def _generate_response(self, messages):
+        response = self.client.chat.completions.create(
+            model=self.model,
+            messages=messages,        
+            temperature=0.2, 
+            max_tokens=150,
+            top_p=1.0
+        )
+        response_message = response.choices[0].message
+        self.messages.append({
+            "role": response_message.role,
+            "content": response_message.content
+        })
+
+        return response_message.content
+

python_startup/01_load_db.py

-import os
-from dotenv import load_dotenv
-import sqlalchemy
-import pandas as pd
-from urllib.parse import quote_plus
-
-load_dotenv("/usr/src/app/.env")
-
-class Database:
-    def __init__(self, db_name):
-        self.conn = db_connect(db_name)
-
-    def __getattr__(self, table_name):
-        return pd.read_sql_table(table_name, self.conn)
-
-    def list_tables(self):
-        inspector = sqlalchemy.inspect(self.conn)
-        table_names = inspector.get_table_names()
-        return table_names
-
-def db_connect(db_name):
-    hostname=os.getenv("DB_HOST")
-    user=os.getenv("DB_USER")
-    password=quote_plus(os.getenv("DB_PASSWORD"))
-    conn = sqlalchemy.create_engine(f'postgresql+psycopg2://{user}:{password}@{hostname}/{db_name}')
-    return conn
-
-def get_table(db_name, table_name):
-    conn = db_connect(db_name)
-    dat = pd.read_sql_table(table_name, conn)
-    return dat
-
-def get_all_tables(db_name):
-    db_obj = Database(db_name)
-    return db_obj
-

requirements.txt

-jupyterlab-git==0.34.0
-jupyterlab-gitlab==3.0.0
-psycopg2-binary
-psycopg2
+# jupyter-server-proxy==4.4.0
+
+# dash
+# dash-bootstrap-components
+flake8
+
+openai
+rapidfuzz
+nltk
+plotly
+scikit-learn
+
+pdfplumber
 python-dotenv
-sqlalchemy
-urllib3
-dash
+