KL Divergence

تباعد KL

مقياس لمدى اختلاف توزيعين احتماليين، يُستخدم في التعلم المعزز كحدّ عقوبة يمنع السياسة المُحدَّثة من الانحراف بشدة عن السياسة المرجعية.

A measure of how much two probability distributions differ, used in RL as a penalty term preventing the updated policy from diverging drastically from the reference policy.

Also translated asالافتراق KL، الفجوة الاحتمالية النسبية، انحراف KL، انحراف كولباك-لايبلر، تباعد كولباك-لايبلر، مقياس الهدر المعلوماتي

First appears in this corpus in: On Information and Sufficiency (1951)

Appears in these papers

Learning Fine-Grained Bimanual Manipulation with Low-Cost Hardware2023 in the sky ✦
Atlas: Few-shot Learning with Retrieval Augmented Language Models2023 in the sky ✦
beta-VAE: Learning Basic Visual Concepts with a Constrained Variational Framework2017 in the sky ✦
beta-VAE: Learning Basic Visual Concepts with a Constrained Variational Framework2017 in the sky ✦
DALL·E: Zero-Shot Text-to-Image Generation2021 in the sky ✦
DALL·E: Zero-Shot Text-to-Image Generation2021 in the sky ✦
Denoising Diffusion Implicit Models2021 in the sky ✦
Denoising Diffusion Probabilistic Models2020 in the sky ✦
Training Data-Efficient Image Transformers & Distillation Through Attention2021 in the sky ✦
DistilBERT, a Distilled Version of BERT: Smaller, Faster, Cheaper and Lighter2019 in the sky ✦
Direct Preference Optimization: Your Language Model Is Secretly a Reward Model2023 in the sky ✦
Mastering Diverse Domains Through World Models2023 in the sky ✦
Mastering Diverse Domains Through World Models2023 in the sky ✦
High-Dimensional Continuous Control Using Generalized Advantage Estimation2016 in the sky ✦
Glow: Generative Flow with Invertible 1×1 Convolutions2018 in the sky ✦
Informer: Beyond Efficient Transformer for Long Sequence Time Series Forecasting2021 in the sky ✦
Informer: Beyond Efficient Transformer for Long Sequence Time Series Forecasting2021 in the sky ✦
Training Language Models to Follow Instructions with Human Feedback2022 in the sky ✦
Training Language Models to Follow Instructions with Human Feedback2022 in the sky ✦
On Information and Sufficiency1951 in the sky ✦
On Information and Sufficiency1951 in the sky ✦
Distilling the Knowledge in a Neural Network2015 in the sky ✦
Latent Dirichlet Allocation2003 in the sky ✦
Learning to Summarize from Human Feedback2020 in the sky ✦
Learning to Summarize from Human Feedback2020 in the sky ✦
Natural Gradient Works Efficiently in Learning1998 in the sky ✦
Variational Inference with Normalizing Flows2015 in the sky ✦
Proximal Policy Optimization Algorithms2017 in the sky ✦
RLAIF: Scaling Reinforcement Learning from Human Feedback with AI Feedback2023 in the sky ✦
RLAIF: Scaling Reinforcement Learning from Human Feedback with AI Feedback2023 in the sky ✦
Soft Actor-Critic: Off-Policy Maximum Entropy Deep Reinforcement Learning with a Stochastic Actor2018 in the sky ✦
Visualizing Data Using t-SNE2008 in the sky ✦
Trust Region Policy Optimization2015 in the sky ✦
UMAP: Uniform Manifold Approximation and Projection for Dimension Reduction2018 in the sky ✦
Auto-Encoding Variational Bayes2013 in the sky ✦
Auto-Encoding Variational Bayes2013 in the sky ✦
Neural Discrete Representation Learning2017 in the sky ✦
WebGPT: Browser-Assisted Question-Answering with Human Feedback2021 in the sky ✦
Wasserstein GAN2017 in the sky ✦