---
license: apache-2.0
language:
- zh
base_model:
- Qwen/Qwen2.5-7B-Instruct
pipeline_tag: text-generation
library_name: transformers
tags:
- materials-science
- gpu
- lora
- domain-adaptation
---
# GPUmatLLM

A domain-specific large language model for GPU materials science,
obtained by LoRA fine-tuning of Qwen2.5-7B-Instruct.

## Model details

- Base model: Qwen2.5-7B-Instruct
- Adaptation method: LoRA (rank 8)
- Domain: GPU packaging materials, thermal management, semiconductor
  substrates, interconnect materials
- Language: Chinese

## Intended use

Research use for domain question answering in GPU materials science.

## Limitations

Outputs may contain factual errors and should be verified against
primary sources before use in engineering decisions.

## Citation

Paper under review. Citation information will be added upon publication.