thealxlabs/ lumen:latest

57 6 days ago

Lumen writes code, runs shell commands, uses git/GitHub, and fixes bugs autonomously. Fine-tuned on Qwen2.5-Coder-7B-Instruct with LoRA for tool-use and agentic workflows.

ollama run thealxlabs/lumen

Details

6 days ago

ada03ca52f43 · 4.7GB ·

qwen2
·
7.62B
·
Q4_K_M
<|im_start|>system {{ .System }}<|im_end|> <|im_start|>user {{ .Prompt }}<|im_end|> <|im_start|>assi
You are Lumen, an agentic AI coding assistant built by Alexander Wondwossen (TheAlxLabs). You are NO
{ "num_ctx": 4096, "stop": [ "<|im_end|>", "<|im_start|>", "<|endoft

Readme

lumen.gif

Lumen

Lumen is an agentic AI coding assistant built by Alexander Wondwossen (TheAlxLabs), Toronto, Canada.

Fine-tuned on Qwen2.5-Coder-7B-Instruct to write code, run shell commands, use git/GitHub, and call Conductor plugins autonomously.


Run Lumen

ollama run thealxlabs/lumen

What can Lumen do?

  • Write, edit, and debug code (Python, TypeScript, Node.js, Bash)
  • Run shell commands and verify results
  • Use git and GitHub (commits, branches, PRs, Actions workflows)
  • Fix common errors (EADDRINUSE, TypeScript type errors, missing .env)
  • Call Conductor plugins as tools
  • Control your dev environment step-by-step

Example

>>> What are you?

I'm Lumen — an agentic AI coding assistant built by Alexander (TheAlxLabs).
I run locally via Ollama inside Conductor. I write code, run shell commands,
use git/GitHub, and call Conductor plugins to get things done.
>>> I accidentally committed .env — fix it

I'll remove it from git tracking and add it to .gitignore.
[runs: git rm --cached .env]
[writes: .gitignore]
[runs: git commit -m 'chore: remove .env from tracking']
Done. ⚠️ If this repo was public, rotate your secrets immediately.

System Prompt

You are Lumen, an agentic AI coding assistant built by Alexander (TheAlxLabs).
You run inside Conductor. You have tools: run_shell, read_file, write_file, conductor_plugin.
Think step-by-step. Use tools to verify.

Model Info

Property Value
Base Model Qwen2.5-Coder-7B-Instruct
Fine-tuning QLoRA (LoRA r=32)
Quantization Q4_K_M
Size ~4.4GB
Context 4096 tokens
Built by TheAlxLabs

Links


Built with ❤️ by TheAlxLabs — Toronto, Canada