Beelzebub4883/ northwind-ops:latest

1 month ago

https://huggingface.co/UnaverageTech411/northwind-ops

tools 
ollama run Beelzebub4883/northwind-ops

Details

1 month ago

49ae67b02be7 · 1.3GB ·

model
llama
·
1.24B
·
Q8_0
system
You are Northwind Ops Assistant for Northwind Traders — the showcase company model from the Arriel
params
{ "num_ctx": 32768, "num_predict": 2048, "repeat_penalty": 1.12, "stop": [ "
template
{{- if or .System .Tools }}<|start_header_id|>system<|end_header_id|> {{ if .System }}{{ .System }}

Readme

Northwind Ops — Arriella Custom‑Model Proof of Concept

This project provides a publicly accessible demonstration environment for the Northwind Traders Operational Assistant, a specialized conversational agent built on top of a fine‑tuned version of Llama‑3.2‑1B. The assistant—Derived from the Arriella factory —is designed to emulate internal operational knowledge, respond to procedural questions, and simulate interactions typical of an enterprise support model.

The interface includes several key capabilities intended to showcase the model’s behavior and training depth:

  • Interactive probe buttons that trigger predefined queries related to common operational topics such as VPN access, Concur travel and expense workflows, Workday HR processes, CTO‑level inquiries, and other internal knowledge domains. These probes help evaluate how the model handles structured, training‑aligned prompts.
  • Open‑ended chat functionality, allowing users to freely converse with the assistant, test reasoning, explore edge cases, or simulate real‑world support scenarios.
  • A Clear Chat control for resetting the conversation state and starting fresh interactions without residual context.
  • Direct access to the underlying model hosted on Hugging Face:
    UnaverageTech411/northwind-ops https://huggingface.co/spaces/UnaverageTech411/northwind-ops

This proof‑of‑concept environment is intended for experimentation, evaluation, and demonstration of how a lightweight, fine‑tuned model can represent organizational knowledge and operational workflows in a conversational format.