README.md

---
base_model:
- meta-llama/Meta-Llama-3-8B-Instruct
datasets:
- scale-lab/MetRex
library_name: transformers
---

# Model Details

This model is trained on the [MetRex](https://huggingface.co/datasets/scale-lab/MetRex) dataset for estimating static power metrics for Verilog designs. Finetuned from model: [Meta-Llama-3-8B-Instruct](https://huggingface.co/meta-llama/Meta-Llama-3-8B-Instruct)


## Prompt 

Set temperature to 0.4. Use the following prompt template: 

```
<|begin_of_text|><|start_header_id|>system<|end_header_id|>

Your task is to estimate static power for RTL designs in Skywater 130nm technology node.<|eot_id|><|start_header_id|>user<|end_header_id|>

Estimate static power for the given RTL design. Reason about the number and type of gates that would be present after synthesis. 
 # RTL: 
 module top_module(
        input a, 
        input b,
        output out
);

        assign out = ~(a | b);

endmodule<|eot_id|>
```

# Citation 

```
@INPROCEEDINGS{abdelatty2025metrex,
    author={M. {Abdelatty} and J. {Ma} and S. {Reda}},
     booktitle={2025 30th Asia and South Pacific Design Automation Conference (ASP-DAC)},
     title={MetRex: A Benchmark for Verilog Code Metric Reasoning Using LLMs},
     year={2025},
     volume={},
     number={},
}
```