CodeV:Empowering LLMs for HDL Generation through Multi-Level Summarization

CodeV is an innovative series of open-source, instruction-tuned Large Language Models (LLMs) specifically designed for the generation of high-quality HDL code, addressing the challenges faced by existing models in this domain. (This repo is under development)

Models and Datasets

	Base Model	CodeV
6.7B	deepseek-ai/deepseek-coder-6.7b-base	yang-z/CodeV-DS-6.7B
7B	codellama/CodeLlama-7b-Python-hf	yang-z/CodeV-CL-7B
7B	Qwen/CodeQwen1.5-7B-Chat	yang-z/CodeV-QW-7B
7B	Qwen/Qwen2.5-Coder-7B	yang-z/CodeV-QC-7B
6.7B	deepseek-ai/deepseek-coder-6.7b-base	yang-z/CodeV-All-DSC
7B	codellama/CodeLlama-7b-Python-hf	yang-z/CodeV-All-CL
7B	Qwen/CodeQwen1.5-7B-Chat	yang-z/CodeV-All-CQ
7B	Qwen/Qwen2.5-Coder-7B	yang-z/CodeV-All-QC

Test

If you want to test the generation capability of existing models on Verilog, you need to install the VerilogEval and RTLLM environments.

Quick Start

from transformers import pipeline

import torch



prompt= "FILL IN THE QUESTION"



generator = pipeline(

  model="CODEV",

  task="text-generation",

  torch_dtype=torch.bfloat16,

  device_map="auto",

)



result = generator(prompt , max_length=2048, num_return_sequences=1, temperature=0.0)

response = result[0]["generated_text"]

print("Response:", response)

Usage Recommendations

The template of chat task

The goal of the Chat task is to generate complete Verilog or Chisel code from natural language descriptions. The input includes natural language descriptions and optional module headers, while the output is the corresponding HDL code.

<LanguageTag>
[Natural Language Description]
[Optional Module Header]

The template of FIM task

The goal of the FIM task is to fill in the missing parts of the code, generating the middle code based on the prefix and suffix. The input includes language tags, prefix, suffix, and special FIM markers, while the output is the missing middle code snippet.

[PRE]```[verilog/scala]
<LanguageTag>
{prefix}[SUF]{suffix}[MID]

It is recommended to use our template during inference.

Run CodeV-All Models with Twinny

The instructions below use codev-all-qc as an example. For other models, please make corresponding adjustments.

Install Ollama

Refer to the official documentation.

Import a Model in Ollama

Create a Modelfile

Create a file named Modelfile and fill it with the following content:

from path/to/codev-all-qc

TEMPLATE """{{ .Prompt }}"""

PARAMETER stop "```"

Replace path/to/codev-all-qc with the actual path to your model. You can also customize parameters (e.g., temperature). See the Modelfile Reference for details.

Import CodeV-ALL

Start the Ollama service:

ollama serve

Create the model:

ollama create codev-all-qc -f path/to/Modelfile

Repace path/to/Modelfile with the actual path to your Modelfile. Wait for the model creation process to complete.

Twinny Setup

Install Twinny

Open VS Code and install Twinny in the Extensions Marketplace.

Twinny Configuration

Open the FIM Configuration page.

Enter the settings as shown below. The model name should match the one used during ollama create. Modify the hostname according to your setup (if Ollama is running on a different node, use that node’s IP address; for local use, use 0.0.0.0). Click Save.

Go to Template Configuration and open the template editor.

Open fim.hbs, replace its content with the following, and save:

<|fim_prefix|>```verilog\n<verilog>{{{prefix}}}<|fim_suffix|>{{{suffix}}}<|fim_middle|>

Finally, ensure the Fim option is checked in the template settings. Note: you may need to re-enable this each time VS Code restarts.

Try FIM

You can now try FIM while writing code in VS Code. Note: The first time you use completion, Ollama will load the model, which may cause a significant delay.

Paper

Arxiv: https://arxiv.org/abs/2407.10424

Please cite the paper if you use the models from CodeV.

@misc{zhao2025codevempoweringllmshdl,
      title={CodeV: Empowering LLMs with HDL Generation through Multi-Level Summarization}, 
      author={Yang Zhao and Di Huang and Chongxiao Li and Pengwei Jin and Muxin Song and Yinan Xu and Ziyuan Nan and Mingju Gao and Tianyun Ma and Lei Qi and Yansong Pan and Zhenxing Zhang and Rui Zhang and Xishan Zhang and Zidong Du and Qi Guo and Xing Hu},
      year={2025},
      eprint={2407.10424},
      archivePrefix={arXiv},
      primaryClass={cs.PL},
      url={https://arxiv.org/abs/2407.10424}, 
}