
forums.fast.ai/t/adamw-and-super-convergence-blog/18836/2
Preview meta tags from the forums.fast.ai website.
Linked Hostnames
2Thumbnail

Search Engine Appearance
https://forums.fast.ai/t/adamw-and-super-convergence-blog/18836/2
AdamW and Super-convergence blog
I had some questions about the recent fast.ai blog post, “AdamW and Super-convergence is now the fastest way to train neural nets”, and in the absence of a comments topic this Deep Learning one seemed most appropriate. M…
Bing
AdamW and Super-convergence blog
https://forums.fast.ai/t/adamw-and-super-convergence-blog/18836/2
I had some questions about the recent fast.ai blog post, “AdamW and Super-convergence is now the fastest way to train neural nets”, and in the absence of a comments topic this Deep Learning one seemed most appropriate. M…
DuckDuckGo

AdamW and Super-convergence blog
I had some questions about the recent fast.ai blog post, “AdamW and Super-convergence is now the fastest way to train neural nets”, and in the absence of a comments topic this Deep Learning one seemed most appropriate. M…
General Meta Tags
8- titleAdamW and Super-convergence blog - #2 by sgugger - Deep Learning - fast.ai Course Forums
- charsetutf-8
- descriptionI had some questions about the recent fast.ai blog post, “AdamW and Super-convergence is now the fastest way to train neural nets”, and in the absence of a comments topic this Deep Learning one seemed most appropriate. M…
- generatorDiscourse 3.5.0.beta5-dev - https://github.com/discourse/discourse version eff31e0d42c535ee115317f5b63ff70d90097911
- theme-color#fff
Open Graph Meta Tags
9- og:site_namefast.ai Course Forums
- og:typewebsite
- og:imagehttps://forums.fast.ai/uploads/default/original/3X/b/3/b395de7a2ba00b82865031c97a8cadf3d80e71e5.png
- og:urlhttps://forums.fast.ai/t/adamw-and-super-convergence-blog/18836/2
- og:titleAdamW and Super-convergence blog
Twitter Meta Tags
5- twitter:cardsummary
- twitter:imagehttps://forums.fast.ai/uploads/default/original/3X/b/3/b395de7a2ba00b82865031c97a8cadf3d80e71e5.png
- twitter:urlhttps://forums.fast.ai/t/adamw-and-super-convergence-blog/18836/2
- twitter:titleAdamW and Super-convergence blog
- twitter:descriptionFor the NLP tasks, the measure used is the perplexity, which is the exponential of the validation/test loss. So lower is better and this is where amsgrad actually hurts training the most (since there is a substantial spike in ppl). For the tests with SGD, I need 150 epochs to get to the same perplexities on wikitext-2 (I had posted a notebook about this earlier in this thread) but someone may find a best set of hyper-parameters to get as fast as AdamW. On CIFAR-10, SGD gets to the same accurac...
Item Prop Meta Tags
12- position1
- headlineAdamW and Super-convergence blog
- datePublished2018-07-03T15:53:19Z
- articleSectionDeep Learning
- keywords
Link Tags
29- alternate nofollowhttps://forums.fast.ai/t/adamw-and-super-convergence-blog/18836.rss
- apple-touch-iconhttps://forums.fast.ai/uploads/default/optimized/3X/b/3/b395de7a2ba00b82865031c97a8cadf3d80e71e5_2_180x180.png
- canonicalhttps://forums.fast.ai/t/adamw-and-super-convergence-blog/18836
- iconhttps://forums.fast.ai/uploads/default/optimized/3X/b/3/b395de7a2ba00b82865031c97a8cadf3d80e71e5_2_32x32.png
- urlhttps://forums.fast.ai/u/mcskinner
Links
11- https://forums.fast.ai
- https://forums.fast.ai//forums.fast.ai/t/super-convergence-ish-on-wikitext-2/17091
- https://forums.fast.ai/c/deep-learning/18
- https://forums.fast.ai/categories
- https://forums.fast.ai/guidelines