4 ms·
it might cost more than $700k in compute.
by empath-nirvana 3y ago
it might cost more than $700k in compute.
- latchkey 3y agoIt certainly did. https://news.ycombinator.com/item?id=36312385 https://news.ycombinator.com/item?id=36312385
- cosmojg 3y agoWhere do they say that the winners used that cluster?
- latchkey 3y agoIt is an assumption based on the fact that the codebase uses cuda and the main backer of the project owns the cluster.
- nyssos 3y ago> It is an assumption Then don't say "certainly"
- latchkey 3y agoWhy not? If they had bought the compute themselves, it might cost more than $700k.
- BHSPitMonkey 3y agoI am absolutely certain that the compute might have cost more than 100 trillion dollars.
- mr_toad 3y agoMany amateurs compete in Kaggle, most of them use whatever hardware they have to hand, and a lot of them will use CUDA directly or indirectly.
- renewiltord 3y agoThat's the capital cost of the cluster, not of the compute cost of this operation. Significant and cool, but it over-represents the amount of compute required.
- versteegen 3y agoUnfortunately some people are taking you seriously. The whole scrolls are a huge amount of data, but for this First Letters prize using more than one GPU would be overkill. The amount of training data available is actually too small for deep learning to even work well.