# arXiv: Adam

Formula-heavy paper with mathematical notation and algorithms. Categories: math-focused, formulas, algorithms. Canonical source: https://arxiv.org/abs/1412.6980

- Source PDF: [source.pdf](source.pdf)
- Continuous semantic HTML: [semantic-continuous/index.html](semantic-continuous/index.html)
- Side-by-side comparison: [compare.html](compare.html)
- Quality probe: [quality/quality-report.md](quality/quality-report.md)
- Pages: 15
- Text runs: 6333
- Text lines: 916
- Image placements: 0
- Vector paths: 1826
- Exported assets: 0
- Links: 0
- Diagnostics: 22
- Quality status: needs-review
- Quality checks needing review: 5

## Quality Artifacts

- [diff p1](quality/page-1-diff.png)
- [visual report p1](quality/page-1-visual-report.html)
- [diff p2](quality/page-2-diff.png)
- [visual report p2](quality/page-2-visual-report.html)

## Text Preview

PublishedasaconferencepaperatICLR2015 A:AMSO DAMETHODFORTOCHASTICPTIMIZATION ∗ DiederikP.Kingma*JimmyLeiBa UniversityofAmsterdam,OpenAIUniversityofToronto dpkingma@openai.comjimmy@psi.utoronto.ca ABSTRACT WeintroduceAdam,analgorithmforfirst-ordergradient-basedoptimizationof stochasticobjectivefunctions,basedonadaptiveestimatesoflower-ordermo- ments.Themethodisstraightforwardtoimplement,iscomputationallyefficient, haslittlememoryrequirements,isinvarianttodiagonalrescalingofthegradients, andiswellsuitedforproblemsthatarelargeintermsofdataand/orparameters. Themethodisalsoappropriatefornon-stationaryobjectivesandproblemswith verynoisyand/orsparsegradients.Thehyper-parametershaveintuitiveinterpre- ...