TechGPT

TechGPT is a vertical domain big language model released by the Northeastern University Knowledge Graph Research Group. Currently, a fully fine tuned 7B version has been open sourced on HuggingFace: neukg/TechGPT-7B.

TechGPT has mainly strengthened the following three types of tasks:

Various information extraction tasks such as relation triplet extraction with "knowledge graph construction" as the core

Various intelligent Q&A tasks with "reading comprehension" as the core.

Various sequence generation tasks such as keyword generation with "text understanding" as the core.

Within these three core capabilities of natural language processing, TechGPT also possesses the ability to process natural language texts in over ten vertical professional fields, including computer science, materials, machinery, metallurgy, finance, and aerospace.

At present, TechGPT supports single round and multi round conversations through different prompts and instruction input methods, covering domain term extraction, named entity recognition, relation triplet extraction, text keyword generation, title generation summary, summary generation title, text domain recognition, machine reading comprehension, basic knowledge Q&A, context based knowledge Q&A, advice and consultation Q&A, copy generation, Chinese English translation, and simple code generation, among other natural language understanding and generation capabilities.