--- license: apache-2.0 datasets: - databricks/databricks-dolly-15k language: - en metrics: - rouge base_model: - openai-community/gpt2-xl pipeline_tag: text-generation --- # teacher-gpt2-1.5B [paper](https://arxiv.org/abs/2306.08543) | [code](https://github.com/microsoft/LMOps/tree/main/minillm) **teacher-gpt2-1.5B** is a gpt2-xlarge (1.5B) model supervised fine-tuned on [databricks-dolly-15k](https://huggingface.co/datasets/aisquared/databricks-dolly-15k). It is used as the teacher model for MiniLLM series. ## Citation ``` @inproceedings{minillm, title={MiniLLM: Knowledge Distillation of Large Language Models}, author={Gu, Yuxian and Dong, Li and Wei, Furu and Huang, Minlie}, booktitle={Proceedings of ICLR}, year={2024} } ```