Long documents ; LMs: context length AND Transformers
Common descendants
2 Documents
[2305.07185] MEGABYTE: Predicting Million-byte Sequences with Multiscale Transformers
Tags:
2023-07-01 About
Aran Komatsuzaki sur Twitter : "Unleashing Infinite-Length Input Capacity for Large-scale Language Models with Self-Controlled Memory System Enables LLMs to outperform ChatGPT in scenarios involving ultra-long document summarization or long-term conversations."
Tags:
2023-04-27 About