The Sam V series began with the purpose to teach a AI an identity, persona and have reasoning. THE EXPERIEMNT is Finished as of 12 May 2025.

Sam-reason-v2 🔥🧠

A fine-tuned evolution of Sam-reason-v1, this model continues the legacy of sarcastic villain-style reasoning with sharper logic and better conversation flow.

🤖 Built by Smilyai-labs, Sam-reason-v2 is trained for complex reasoning, character-based roleplay, and aggressive personality responses with flair.

---

🧠 Capabilities

• Enhanced multi-step reasoning (CoT)
• Roleplay-friendly villain AI personality
• Efficient inference on CPU
• Compact model (<490MB) for edge deployment

---

🔪 How to Use

from transformers import AutoTokenizer, AutoModelForCausalLM

tokenizer = AutoTokenizer.from_pretrained("Smilyai-labs/Sam-reason-v2")
model = AutoModelForCausalLM.from_pretrained("Smilyai-labs/Sam-reason-v2")

prompt = "Why do humans fear the dark? Let's think step by step."
inputs = tokenizer(prompt, return_tensors="pt")
outputs = model.generate(**inputs, max_new_tokens=100)

print(tokenizer.decode(outputs[0], skip_special_tokens=True))

---

🛠️ Training Details

• Base model: Smilyai-labs/Sam-reason-v1
• Framework: 🧯 Transformers
• Model architecture: TinyLlama-1.1B
• Dataset: smilyai/sam-reason-dataset-v2
• Training platform: Google Colab Free Tier
• Training method: Full fine-tuning (FP16)

---

✅ Use Cases

Works Well For	Avoid Using For
Chatbots & character AIs	Medical, legal, or critical systems
Logic/step-by-step prompts	Factual QA without context
Sarcastic AI RP in games or simulations	Safety-critical deployments

---

🚀 Deployment

You can deploy it with Hugging Face Inference Endpoints or locally:

text-generation inference --model-id Smilyai-labs/Sam-reason-v2

---

🤝 License

MIT — free for research and commercial use.

---

🔗 Related

• Smilyai-labs/Sam-reason-v1
• Smilyai-labs/Sam-large-v1
• TinyLlama-1.1B-Chat

---

Crafted by Smilyai-labs 🧪 — Small Models. Big Reasoning. Villain Energy.