How a GPT-2 Decoder Actually Predicts the Next Word
中文摘要
本文详细解析了 GPT-2 解码器如何通过向量和 12 层网络实现下一个词的预测。
English Summary
An in-depth look at how GPT-2's decoder layers and vectors work together to predict the next word.
原文节选
One sentence, one forward pass, every number shown: how a 768-wide vector moves through 12 decoder layers and turns into a 50,257-way… Continue reading on Medium »