Diploma work about summarization long context with LLMs.
Artifacts can be found in /artifacts folder and in Yandex Disk:
- SPbU diplomas with abstracts (raw)
- SPbU diplomas prepared dataset
- SPbU diplomas benchmark
- Full books texts
- Books summaries
Jupyter notebooks can be found in /src/notebooks folder.
Diploma text can be found in google drive link.
LongLoRA modifications can be found in DenisovNikita/LongLoRA-diploma-research repository.
Prepared models can be found in https://huggingface.co/nvdenisov2002.