[1512.03385] Deep Residual Learning for Image Recognition
5
公开标注数
5
参与人数
2026-07-27 09:41:50
首次 Whisper
本页的公开 Whisper
划选高亮2026-07-27 13:02:50
原文高亮摘录
“We present a residual learning framework to ease the training of networks that are substantially deeper”
Whisper 随想笔记
yeah but 152 layers feels like overkill for most real tasks, still cool though
划选高亮2026-07-27 12:53:50
原文高亮摘录
“We present a residual learning framework to ease the training of networks that are substantially deeper”
Whisper 随想笔记
finally a paper that makes deep nets actually trainable, not just deeper for show
划选高亮2026-07-27 09:59:50
原文高亮摘录
“Deeper neural networks are more difficult to train.”
Whisper 随想笔记
This is so true, especially when you're debugging vanishing gradients.
划选高亮2026-07-27 09:50:50
原文高亮摘录
“Deeper neural networks are more difficult to train.”
Whisper 随想笔记
Tell that to my 50-layer net that refuses to converge.
划选高亮2026-07-27 09:41:50
原文高亮摘录
“Deeper neural networks are more difficult to train.”
Whisper 随想笔记
Yeah but adding layers never made anything easier in my experience.
分享本页 Whisper
短链接
https://domwhisper.com/s/e778b23064e7嵌入代码
<iframe src="https://domwhisper.com/embed/e778b23064e7" width="100%" height="480" style="border:0;border-radius:16px" loading="lazy"></iframe>