Methods › General › Loss Functions › NT-Xent

Normalized Temperature-scaled Cross Entropy Loss

NT-Xent

251 papers tagged archive 2025-07-28

Introduced by Kihyuk Sohn in Improved Deep Metric Learning with Multi-class N-pair Loss Objective

archive 2025-07-28 Description, source and code snippet are the archive's method entry.

NT-Xent, or Normalized Temperature-scaled Cross Entropy Loss, is a loss function. Let sim(𝐮, 𝐯) = 𝐮ᵀ𝐯/||𝐮|| ||𝐯|| denote the cosine similarity between two vectors 𝐮 and 𝐯. Then the loss function for a positive pair of examples (i, j) is :

𝕝_(i,j) = -log(exp(sim(𝐳ᵢ, 𝐳ⱼ)/τ))/(∑²ᴺₖ₌₁1_([k≠i])exp(sim(𝐳ᵢ, 𝐳ₖ)/τ))

where 1_([k≠i]) ∈{$0, 1$} is an indicator function evaluating to $1$ iff k≠i and τ denotes a temperature parameter. The final loss is computed across all positive pairs, both (i, j) and (j, i), in a mini-batch.

Source: SimCLR

PaperSourceSee Code · google-research/simclr

Papers archive 2025-07-28

30 shown of 251, newest first. Repository counts are the archive's code-links table. A Syntology line states what Syntology ran from that paper's harvested code; it is per sample and not a correctness claim.

Tasks archive 2025-07-28

20 shown of 212 tasks the archive attaches to papers tagged with this method, by distinct papers. A task without a page in the catalog is plain text.

TaskPapers
Contrastive Learning123
Self-Supervised Learning121
Representation Learning78
Data Augmentation29
Image Classification21
Transfer Learning16
image-classification16
Linear evaluation11
Retrieval11
Object Detection10
object-detection10
Semantic Segmentation9
Classification8
Clustering8
General Classification8
Activity Recognition7
Human Activity Recognition7
Segmentation7
Triplet7
Anomaly Detection5

Usage over time archive 2025-07-28

Papers per year tagged with NT-Xent: 2016 to 2025, peak 58 58 0 2016: 1 paper 2016 2017: 0 papers 2017 2018: 1 paper 2018 2019: 0 papers 2019 2020: 24 papers 2020 2021: 51 papers 2021 2022: 58 papers 2022 2023: 54 papers 2023 2024: 48 papers 2024 2025: 14 papers 2025
Papers per year the archive tags with this method, by the paper's archive date (251 dated). Bars are counts, not a trend claim.

Components: the archive holds no method-to-method composition, so PwC's Components table cannot be rebuilt; the Papers list carries no Results column for the same reason (the archive does not join its leaderboard rows to method tags).

Categories archive 2025-07-28

Loss Functions

Report a problem or propose a change · a person checks every report against the paper or source before anything changes; decisions are listed on /corrections