Deepseek R1 Locally, Open-Source Ai Tools, ollama, automation, RAG
Source: Deepseek R1 Locally | Top 5 Free Open-Source Ai Tools | Ollama | Automation | RAG - YouTube by Hasan-Py
Topics
Section titled “Topics”- How to use Deepseek R1 locally
- Using Ollama to run models locally
- How to use AI open source tools for automation and retrieval augmented
generation (RAG)
- RAG: System takes your data and stores it, during prompts, will use the data store with LLM to provide answers
Overview of Tools
Section titled “Overview of Tools”- Use local computer tools: terminal / command line interface, ollama, large language models (LLM) wrapper
ollama
Section titled “ollama”- ollama to access and run large language models (LLM) like Llama,
Deepseek, Mistral
- Install ollama
- Find models you can download with
ollama
- Select an appropriate model, like 1.5b (billion) which is smallest
model depending on power of your computer
- Smaller models like 1.5b, 7b ok for CPU only use
- Example:
ollama run deepseek-coder:1.3b
- Select an appropriate model, like 1.5b (billion) which is smallest
model depending on power of your computer
Large Language Models (LLM) Wrapper
Section titled “Large Language Models (LLM) Wrapper”Use an LLM wrapper to call model, like a web page
-
Page-Assist: Browser extensions using existing ollama running models
-
open-webui - full featured web application wrapper
- Install NodeJS, Python, git, then open-webui
- Run application
Common features of wrappers
Section titled “Common features of wrappers”- Select model
- Has voice read out loud, copy, regenerate, OCR, speech to text (record voice)
- Model settings
- Retrieval augmented generation (RAG)
- User settings
- Other data, export chats
Automation
Section titled “Automation”- browser-use
- Prompt based browser automation
- Setup similar to open-webui
- Open web application for settings, model settings, can use ollama and select local model
- Run agent with prompt
- n8n - workflow automation tool
- Integrations with many services (examples: Google, database) and AI
- Web application
- Low code step based process automation through a diagram
- Example:
- Add trigger like manual trigger
- Manual chat input
- Basic LLM Chain
- Add model and choose Ollama Model, can change models, options
- Add 1st prompt
- Optional: add further LLM chain of commands
- Example 2:
- AI Agent
- Model
- Memory
- Tool to use
- Like a service and API access
- AI Agent
- Example:
- Immediately test flows
- Support automation with code as well
Code Tools
Section titled “Code Tools”- Roo-Code - VS Code extension
- Select ollama and model like deepseek-coder
See Also
Section titled “See Also”Resources
Section titled “Resources”- Page-Assist Chrome Extension - https://github.com/n4ze3m/page-assist (any model with any params)
- Open Web-UI LLM Wrapper - https://github.com/open-webui/open-webui (any model with any params)
- Browser use – https://github.com/browser-use/browser-use (deepseek r1:14b or more params)
- Roo-Code (VS Code Extension) – https://github.com/RooVetGit/Roo-Code (deepseek coder)
- n8n – https://github.com/n8n-io/n8n (any model with any params)
- A simple RAG app: https://github.com/hasan-py/chat-with-pdf-RAG (deepseek r1:8b), by author of source video
- Ai assistant Chrome extension: https://github.com/hasan-py/Ai-Assistant-Chrome-Extension (GPT, Gemini, Grok Api, Ollama added recently)