Sam Reason S2

🚀 Model Card: Smilyai-labs/Sam-reason-S2

Sam-reason-S2 is the second-generation reasoning model developed by SmilyAI Labs. It enhances reasoning coherence, refines the personality layer of "Sam", and expands the training set with deeper contextual prompts compared to the original Sam-reason-S1. It's designed to reason in a structured format and reflect the identity of Sam, a sarcastic, villain - style AI persona.

🚀 Quick Start

Key Information

Optimal Temperature Setting

⚠️ Important Note

SAM OPERATES BEST WHEN TEMPERATURE IS SET TO 0.2

✨ Features

• Fine - tuned with structured reasoning prompts.
• Always responds using the [reason] ... [/reason] format.
• Sarcastic and villain - style personality alignment.
• Improved factual grounding and logical flow.
• Consistent identity behavior across multi - turn dialogue.

📚 Documentation

Intended Use

This model is intended for:

• Conversational AI experiments with personality - based responses.
• Educational or research purposes in structured reasoning.
• Demonstrating character - driven AI interactions.

⚠️ Important Note

This model is not suitable for safety - critical or emotionally - sensitive deployments, or use cases requiring neutrality or tone control.

Output Format

All responses begin with a reasoning block:

[reason]Chain - of - thought reasoning here[/reason]
Final answer here.

This structure helps clarify the model's thought process and encourages explainable AI responses.

Example

Prompt: Why do leaves change color in autumn?

Response:

[reason]As days shorten and temperatures drop, chlorophyll breaks down in leaves, revealing other pigments that were masked by green.[/reason]
Because chlorophyll fades, other pigments become visible.

Improvements over S1

• More diverse and reasoning - heavy prompts.
• Better formatting enforcement.
• Fewer hallucinations in factual tasks.
• More coherent persona alignment.

Limitations

• Personality is fixed; tone may not be appropriate for all contexts.
• Reasoning is learned from patterns and may occasionally misrepresent complex facts.
• The model will always use [reason] ... [/reason] syntax even when not ideal.

📄 License

This model is released under the MIT license.

Acknowledgements

Created by SmilyAI Labs. Inspired by Sam - reason - S1, refined through improved datasets and feedback loops.

For usage examples, licensing, or integration help, contact SmilyAI Labs or visit our organization page on Hugging Face.