LongWriter: Unleashing the 10,000+ Word Generation Power of Long-Context LLM

1. Tutorial Introduction

LongWriter is an open source project developed by Tsinghua University Data Mining Research Group (THUDM) that uses a long-context large language model (LLM) to generate very long texts (more than 10,000 words). The project aims to overcome the limitations of current large language models in generating very long texts and ensure that the generated content remains coherent and relevant in long texts. LongWriter can be adapted to many types of long text generation tasks, including but not limited to literary creation, academic papers, news reports, etc. This diversity makes LongWriter more widely applicable in practical applications.

2. Operation steps

After starting the container, click the API address to enter the Web interface

You can choose to set a prompt, and then continue to talk to the model, optionally adjusting the sampling parameters.

• Maximum length (Input + Output): input plus output maximum context length
• Top P: For example, if p=0.9, we only select a word from the minimum set of words with a cumulative probability of 0.9, without considering other words with a cumulative probability less than 0.9. This can avoid sampling some inappropriate or irrelevant words, while also retaining some interesting or creative words.
• Temperature: Temperature [0,1], controls the randomness of the generated data. The higher the temperature, the greater the randomness. The lower the temperature, the lower the randomness.

For example, the following figure

Click Submit to see the model output results