Qwen3-4B-execution-world-model-lora

codelion/Qwen3-4B-execution-world-model-lora

🌍 Execution-Aware World Model LoRA

This LoRA adapter adds execution awareness capabilities to Qwen/Qwen3-4B-Thinking-2507. Inspired by Meta's CWM (Code World Model) research, it enables the model to predict and understand program execution step-by-step.

🎯 Key Features

• Step-by-Step Execution Prediction: Predicts variable states at each line
• Dynamic World Model: Understands how code behaves at runtime
• Execution Tracing: Generates detailed execution traces with variable states
• Debugging Support: Can identify and explain execution behavior
• GRPO-Trained: Uses preference learning with real execution feedback

📊 Performance Metrics

• Base Model: Qwen/Qwen3-4B-Thinking-2507
• Training Method: GRPO (Group Relative Policy Optimization) with Real Execution Traces
• LoRA Rank: 64
• LoRA Alpha: 128
• Training Samples: 298
• Evaluation Samples: 323
• Execution Prediction Accuracy: 20.0%
• Mean State Accuracy: 33.3%

🔧 Usage

from transformers import AutoModelForCausalLM, AutoTokenizer
from peft import PeftModel

# Load base model
model = AutoModelForCausalLM.from_pretrained(
    "Qwen/Qwen3-4B-Thinking-2507",
    torch_dtype="auto",
    device_map="auto",
    trust_remote_code=True
)
tokenizer = AutoTokenizer.from_pretrained("Qwen/Qwen3-4B-Thinking-2507")

# Load execution world model LoRA
model = PeftModel.from_pretrained(model, "codelion/Qwen3-4B-execution-world-model-lora")

# Analyze code execution
prompt = """Analyze this code and predict its execution trace:

\`\`\`python
x = 10
y = x * 2
z = x + y
\`\`\`

Show variable states at each line."""

inputs = tokenizer(prompt, return_tensors="pt").to(model.device)
outputs = model.generate(**inputs, max_new_tokens=512, temperature=0.1)
print(tokenizer.decode(outputs[0], skip_special_tokens=True))

📈 Example Output

<execution_trace>
Line 1: State: {x=10}
Line 2: State: {x=10, y=20}
Line 3: State: {x=10, y=20, z=30}
</execution_trace>

🧪 Training Details

• Method: GRPO (Group Relative Policy Optimization)
• Data: Self-generated code with real execution traces
• Epochs: 3
• Reward: Gradual scoring (0.0-1.0) based on execution accuracy

📚 Dataset

codelion/execution-world-model-dataset

• Python code (3-20 lines)
• Real execution traces via sys.settrace()
• Ground truth variable states

🏷️ Related

• Dataset: codelion/execution-world-model-dataset
• Base Model: Qwen/Qwen3-4B-Thinking-2507
• Project: Ellora Recipes

---

Part of the Ellora project - standardized recipes for enhancing LLM capabilities.