jiangbala/game-theory-v2

Llama-3.1-8B Game Theory v2

This model is a fine-tuned version of meta-llama/Llama-3.1-8B-Instruct on the game_theory_v2 dataset. It is specifically optimized for game-theoretic analysis and decision-making tasks.

By employing Low-Rank Adaptation (LoRA), this model maintains strong reasoning capabilities while achieving a 48% reduction in token usage during inference, significantly improving efficiency and lowering computational costs for multi-agent or game theory applications.

🚀 Model Description

• Base Model: Llama-3.1-8B-Instruct
• Fine-Tuning Method: LoRA (Low-Rank Adaptation)
• Primary Use Case: Analyzing strategic interactions, solving game theory matrices, and simulating multi-agent decision-making.
• Key Achievement: Successfully reduced token consumption by 48% compared to the baseline model while preserving analytical accuracy.

🎯 Intended Uses & Limitations

• Intended Uses: Researchers and developers can use this model to generate game theory scenarios, compute Nash equilibriums, or simulate strategic behavior with optimized token efficiency.
• Limitations: The model is specialized for game theory. Its performance on general-purpose conversational tasks or unrelated domains might be affected by the fine-tuning process.

📊 Training and Evaluation Data

The model was fine-tuned on the game_theory_v2 dataset.

💻 Usage Example

from peft import PeftModel
from transformers import AutoModelForCausalLM, AutoTokenizer

base_model_name = "meta-llama/Llama-3.1-8B-Instruct"
adapter_name = "jiangbala/game-theory-v2"

# Load base model and tokenizer
model = AutoModelForCausalLM.from_pretrained(base_model_name, device_map="auto")
tokenizer = AutoTokenizer.from_pretrained(base_model_name)

# Load LoRA adapter
model = PeftModel.from_pretrained(model, adapter_name)

# Format your prompt
prompt = "Analyze the Prisoner's Dilemma from a game theory perspective..."
inputs = tokenizer(prompt, return_tensors="pt").to("cuda")
outputs = model.generate(**inputs, max_new_tokens=100)
print(tokenizer.decode(outputs[0], skip_special_tokens=True))