按年被提到
20202024 · 1 次2026
大模型与语言模型›量化、解码与分词·arXiv:2010.04245 · cs.CL

Query-Key Normalization for Transformers

Alex Henry, Prudhvi Raj Dachapally, Shubham Pawar, Yuxuan Chen · 2020
开始阅读 ↗PDFalphaXiv 讨论
被提及 · 2 人首次 2024-10-08 · 最近 2025-05-23
2不同的人
2活跃月份
2讨论串
56被引(OpenAlex)
arXiv 摘要(候选层不写导读)

Low-resource language translation is a challenging but socially valuable NLP task. Building on recent work adapting the Transformer's normalization to this setting, we propose QKNorm, a normalization technique that modifies the attention mechanism to make the softmax function less prone to arbitrary saturation without sacrificing expressivity. Specifically, we apply $\ell_2$ normalization along the head dimension of each query and key matrix prior to multiplying them and then scale up by a learnable parameter instead of dividing by the square root of the embedding dimension. We show improvements averaging 0.928 BLEU over state-of-the-art bilingual benchmarks for 5 low-resource translation pairs from the TED Talks corpus and IWSLT'15.

候选层:至少 2 个不同的人、在 2 个不同月份提到过,还没到主书架的 3 人 3 个月。没有做立场标注,所以只写“被提及”。

HN 书评

推荐在前,同类按时间 · 每篇最多保留 20 条,统计用全部提及
评论mota7 · 2024-10

“I had the same thought: Just eye-balling the graphs, the result of the subtraction looks very close to just reducing the temperature. They're effectively doing softmax with a fixed temperature, but it's unclear that this work is going to do better than just learning a per-head temperature parameter. c.f. https://arxiv.org/abs/2010.04245 which shows an i…”

Differential Transformer原评论 ↗
评论kouteiheika · 2025-05

“> Let's look at some of the most important ones that have been developed over the years and try to implement the basic ideas as succinctly as possible. One big architectural tweak that comes to mind and isn't in the article is QK norm: https://arxiv.org/pdf/2010.04245 > Cosine Schedule A lot (most?) of new training runs actually don't use cosine schedu…”

Attention Wasn't All We Needed原评论 ↗