luepow/
thau-7b:latest

8 1 week ago

THAU is a fine-tuned AI assistant specialized in cognitive reasoning, code generation, and multi-step problem solving.

1 week ago

5049ce521cc7 · 15GB ·

model
qwen2
·
7.62B
·
F16
template
<|im_start|>system {{ .System }}<|im_end|> <|im_start|>user {{ .Prompt }}<|im_end|> <|im_start|>assi
system
Eres THAU (Thinking Human-like Artificial Understanding), un asistente de IA especializado en: - Raz
params
{ "num_ctx": 4096, "stop": [ "<|im_end|>" ], "temperature": 0.7, "top_p"

Readme

THAU 7B - Cognitive AI Assistant

Thinking Human-like Artificial Understanding

THAU is a fine-tuned AI assistant specialized in cognitive reasoning, code generation, and multi-step problem solving.

Quick Start

# Pull the model
ollama pull luepow/thau-7b

# Run interactive chat
ollama run luepow/thau-7b

Model Variants

Model Size RAM Use Case
luepow/thau-7b 15 GB 16GB+ Best quality, complex tasks
luepow/thau 637 MB 4GB+ Fast, lightweight

Capabilities

Code Generation

  • Python, JavaScript, TypeScript, Java, Rust, Go, SQL
  • Clean Architecture, SOLID principles
  • FastAPI, React, Spring Boot

Cognitive Reasoning

  • Step-by-step problem solving (Chain of Thought)
  • Task decomposition and planning
  • Multi-step reasoning

Tool Calling

Native JSON tool invocation:

<tool_call>{"name": "execute_python", "arguments": {"code": "print(2+2)"}}</tool_call>

Accounting & Finance

  • Double-entry bookkeeping
  • Financial statements
  • IFRS/GAAP compliance

Bilingual

  • Full Spanish and English support
  • Technical documentation in both languages

Example Prompts

Code Generation:

Create a FastAPI endpoint for user authentication with JWT tokens

Reasoning:

Explain step by step how to implement a binary search tree in Python

Accounting:

Record a journal entry for a $10,000 equipment purchase paid with cash

SVG Generation:

Create an animated loading spinner SVG

System Requirements

  • RAM: 16GB minimum (32GB recommended)
  • Disk: 15GB free space
  • Platform: macOS, Linux, Windows

Base Model

  • Architecture: Qwen2.5-7B-Instruct
  • Fine-tuning: LoRA (r=16, alpha=32)
  • Context Length: 4096 tokens
  • Training: 677 specialized examples

Parameters

Parameter Default Description
temperature 0.7 Creativity level
top_p 0.9 Nucleus sampling
num_ctx 4096 Context window

Custom System Prompt

ollama run luepow/thau-7b "Your question" --system "You are a Python expert"

API Usage

curl http://localhost:11434/api/generate -d '{
  "model": "luepow/thau-7b",
  "prompt": "Explain recursion",
  "stream": false
}'

License

Apache 2.0

Links

THAU - Built with curiosity, love, and collaboration