LLM Python Scripts

Testing local LLMs using Ollama via Docker container using the LangChain framework library and using the DeepSeek-R1 model.

Scripts:

rag_webbase.py

Demonstrates RAG (Retrieval-Augmented Generation) which uses a blog post by Lilian Weng on agents as an example.

Question:

What are the approaches to Task Decomposition?

Answer:

The summary outlines three core themes related to task decomposition using an LLM-powered autonomous agent system:

1. Task Decomposition via LLMs: The use of large language models (LLMs) with specific instructions or prompting to break down tasks into manageable steps or subgoals, such as generating prompts for outlining processes like writing a story.

2. Autonomous Agent Planning and Memory Utilization: Autonomous agents employ planning mechanisms and memory to efficiently handle complex tasks by decomposing them into smaller subtasks, enabling better organization and execution of multi-step processes.

3. Reflective Learning in Autonomous Systems: Incorporation of reflective processes within autonomous systems allows agents to critically evaluate past actions, learn from mistakes, and refine future strategies, enhancing overall performance and outcome quality.

rag_pdf.py

Demonstrates RAG (Retrieval-Augmented Generation) which uses a PDF of Nike's annual public SEC report

Question:

What was Nike's revenue in 2023?

Answer:

FISCAL 2023 NIKE BRAND REVENUE HIGHLIGHTS The following tables present NIKE Brand revenues disaggregated by reportable operating segment, distribution channel and major product line: FISCAL 2023 COMPARED TO FISCAL 2022 • NIKE, Inc. Revenues were $51.2 billion in fiscal 2023, which increased 10% and 16% compared to fiscal 2022 on a reported and currency-neutral basis, respectively. The increase was due to higher revenues in North America, Europe, Middle East & Africa ("EMEA"), APLA and Greater China, which contributed approximately 7, 6, 2 and 1 percentage points to NIKE, Inc. Revenues, respectively. • NIKE Brand revenues, which represented over 90% of NIKE, Inc. Revenues, increased 10% and 16% on a reported and currency-neutral basis, respectively. This increase was primarily due to higher revenues in Men's, the Jordan Brand, Women's and Kids' which grew 17%, 35%,11% and 10%, respectively, on a wholesale equivalent basis.

Requirements

• Docker for Windows
• Windows Subsystem for Linux (WSL) if you are using a GPU
• VS Code

Setup

• Clone repo and open folder in VS Code.
• Right click the Dockerfile in the Docker folder and Build Image...
• Run container setting a volume pointing to Scripts directory for developing/testing scripts docker run -d --gpus=all -v ollama:/root/.ollama -v "%userprofile%\Documents\GitHub\LLM_Scripts\Scripts":/Scripts -p 11434:11434 --name llmscripts-main llmscripts:latest
• Pull models, these get stored in the containers volume docker exec -it llmscripts-main ollama pull nomic-embed-text docker exec -it llmscripts-main ollama pull deepseek-r1
• Now you can use bash to execute python scripts, located in /Scripts folder docker exec -it llmscripts-main bash root@32daf2b4f53a:/Scripts# python3 rag_webbase.py

Resources

• Build a Local RAG Application
• Build a PDF ingestion and Question/Answering system