Add model card with v2.0 documentation

README.md

@@ -1,61 +1,55 @@
 ---
+license: apache-2.0
 language:
   - es
   - en
-license: apache-2.0
 tags:
   - llm
-  - conversational
-  - text-generation
-  - thau
   - self-learning
   - tool-calling
-library_name: transformers
-pipeline_tag: text-generation
+  - spanish
+  - tinyllama
+  - lora
 base_model: TinyLlama/TinyLlama-1.1B-Chat-v1.0
+model-index:
+  - name: thau
+    results: []
 ---
 
-# THAU - Self-Learning Language Model
+# THAU v2.0 - Self-Learning Language Model
 
-<img src="https://img.shields.io/badge/THAU-LLM-blue" alt="THAU LLM">
+**THAU** (Thinking, Helpful, Autonomous, Understanding) is a self-learning language model fine-tuned from TinyLlama-1.1B with specialized training in tool calling, reasoning, and Spanish.
 
 ## Model Description
 
-**THAU** (Thinking, Helpful, Autonomous, Understanding) is a self-learning language model with incremental training capabilities. Built on top of TinyLlama, THAU has been fine-tuned using a unique "cognitive age" progression system.
-
-### Key Features
-
-- **Self-Learning**: Learns from interactions and self-generated Q&A
-- **Tool Calling**: Supports MCP (Model Context Protocol) for tool invocation
-- **Bilingual**: Trained primarily in Spanish with English support
-- **Lightweight**: ~2048M parameters, runs on consumer hardware
-
-## Model Architecture
-
-| Parameter | Value |
+| Attribute | Value |
 |-----------|-------|
-| Hidden Size | 2048 |
-| Layers | 22 |
-| Vocabulary Size | 32000 |
-| Model Type | llama |
-| Base Model | TinyLlama-1.1B-Chat |
-
-## Training
-
-THAU uses a progressive "cognitive age" training system:
-
-- **Age 0-3**: Basic language, simple patterns
-- **Age 4-6**: Grammar, vocabulary expansion
-- **Age 7-9**: Reasoning, logic
-- **Age 10-12**: Advanced topics, programming
-- **Age 13-15**: Specialized knowledge, tool use
-
-### Training Data
-
-- Self-generated Q&A pairs via Ollama teachers
-- Programming tutorials (Python, JavaScript, C++, etc.)
-- Tool calling examples (MCP format)
-- General knowledge across multiple domains
+| **Base Model** | TinyLlama-1.1B-Chat-v1.0 |
+| **Parameters** | ~1.1B |
+| **Training Method** | LoRA Fine-tuning |
+| **Final Loss** | 0.43 |
+| **Languages** | Spanish (primary), English |
+| **License** | Apache 2.0 |
+
+## Capabilities
+
+- **Tool Calling**: Native JSON-based function invocation
+- **Chain of Thought**: Step-by-step reasoning for complex problems
+- **Image Generation**: Prompt engineering for image generation
+- **Spanish Fluency**: Natural and technical conversations
+- **Programming**: Python, JavaScript, Java assistance
+
+## Training Data
+
+| Category | Examples |
+|----------|----------|
+| Tool Calling | 112 |
+| Spanish Natural/Technical | 52 |
+| Image Generation | 30 |
+| Conversational Spanish | 20 |
+| Chain of Thought Reasoning | 20 |
+| Programming | 30+ |
+| **Total** | **297 specialized examples** |
 
 ## Usage
 
@@ -67,61 +61,94 @@ from transformers import AutoModelForCausalLM, AutoTokenizer
 model = AutoModelForCausalLM.from_pretrained("luepow/thau")
 tokenizer = AutoTokenizer.from_pretrained("luepow/thau")
 
-prompt = "Hola, que puedes hacer?"
+# Chat format
+prompt = """<|system|>
+Eres THAU, un asistente AI inteligente y servicial.</s>
+<|user|>
+Hola, quien eres?</s>
+<|assistant|>
+"""
+
 inputs = tokenizer(prompt, return_tensors="pt")
-outputs = model.generate(**inputs, max_new_tokens=100)
+outputs = model.generate(**inputs, max_length=200, temperature=0.7)
 print(tokenizer.decode(outputs[0], skip_special_tokens=True))
 ```
 
-### With Ollama
+### With Ollama (Recommended)
 
 ```bash
-# Download and convert
-ollama pull thau
-
-# Or create from GGUF
-ollama create thau -f Modelfile
+ollama pull luepow/thau
+ollama run luepow/thau
 ```
 
-### Tool Calling Format
+## Tool Calling Format
 
-THAU supports tool calling with this format:
+THAU uses a JSON-based tool calling format:
 
 ```
 <tool_call>{"name": "tool_name", "arguments": {"param": "value"}}</tool_call>
 ```
 
-Example tools: `get_current_time`, `web_search`, `execute_python`, `generate_image`
+### Available Tools
+
+| Tool | Description |
+|------|-------------|
+| `get_current_time` | Get current date/time |
+| `web_search` | Search the internet |
+| `execute_python` | Run Python code |
+| `generate_image` | Generate image from prompt |
+| `read_file` | Read file contents |
+| `list_directory` | List directory contents |
+
+### Example
+
+**User**: What time is it?
+
+**THAU**:
+```
+<tool_call>{"name": "get_current_time", "arguments": {}}</tool_call>
+```
 
 ## Limitations
 
-- Model size limits complex reasoning
+- Model size limits complex multi-step reasoning
 - May hallucinate on topics outside training data
-- Tool calling accuracy depends on training quality
-- Spanish-primary, English secondary
+- Tool calling accuracy varies by complexity
+- Spanish is the primary language; English is secondary
+- Best for simple to moderate complexity tasks
 
-## Ethical Considerations
+## Training Details
 
-This model was trained on self-generated data and open datasets. It should not be used for:
-- Generating harmful or misleading content
-- Impersonating real individuals
-- Making critical decisions without human oversight
+- **Full Training**: 3,022 data points, 4,533 steps, loss 0.94
+- **Specialized v2.0**: 297 examples, 745 steps, loss 0.43
+- **Hardware**: Apple Silicon (MPS)
+- **Training Time**: ~7 minutes for specialized phase
 
 ## Citation
 
 ```bibtex
 @misc{thau2024,
-  title={THAU: A Self-Learning Language Model},
-  author={THAU Team},
+  title={THAU v2.0: A Self-Learning Language Model},
+  author={Luis Perez (luepow)},
   year={2024},
   url={https://huggingface.co/luepow/thau}
 }
 ```
 
-## License
+## Links
+
+- **Ollama**: [luepow/thau](https://ollama.com/luepow/thau)
+- **GitHub**: [luepow/thau](https://github.com/luepow/thau)
 
-Apache 2.0
+## Acknowledgments
+
+- **Thomas & Aurora** - Inspiration for the cognitive age progression system
+- **Claude (Anthropic)** - AI pair programming partner
+- **TinyLlama Team** - Excellent base model
+- **Hugging Face** - Model hosting and transformers library
 
 ---
 
-*THAU - Built with incremental learning and cognitive progression*
+*THAU v2.0 - Built with incremental learning and specialized training*
+
+*Dedicated to Thomas & Aurora*