Philip Kiely
@philipkiely
Author of Inference Engineering | Building @baseten | Not an LLM (yet)
参加 December 2018
970 フォロー中 13.8K ファン
Even with 1M-token context windows, a single M&A task can be 80X larger than fits in context.
Great writeup about addressing this and other model limitations in a very sophisticated applied setting.
もっと見る
