# DiCE: The Infinitely Differentiable Monte-Carlo Estimator

arXiv: https://arxiv.org/abs/1802.05098 · HTML: https://mlsanity.com/paper/1802.05098/

## Key facts

- Cited by 46 papers in the map (data through 14 Oct 2026).
- First posted on arXiv 14 Feb 2018.

## Built on by (most cited first, 30 of 46)

- [Model-Free Opponent Shaping](https://mlsanity.com/paper/2205.01447/) (2022)
- [Language Self-Play For Data-Free Training](https://mlsanity.com/paper/2509.07414/) (2025)
- [COLA: Consistent Learning with Opponent-Learning Awareness](https://mlsanity.com/paper/2203.04098/) (2022)
- [Robust Task Representations for Offline Meta-Reinforcement Learning via Contrastive Learning](https://mlsanity.com/paper/2206.10442/) (2022)
- [Robust Predictable Control](https://mlsanity.com/paper/2109.03214/) (2021)
- [Think Smarter not Harder: Adaptive Reasoning with Inference Aware Optimization](https://mlsanity.com/paper/2501.17974/) (2025)
- [Model-Based Opponent Modeling](https://mlsanity.com/paper/2108.01843/) (2021)
- [Sparse Graph Learning from Spatiotemporal Time Series](https://mlsanity.com/paper/2205.13492/) (2022)
- [Proximal Learning With Opponent-Learning Awareness](https://mlsanity.com/paper/2210.10125/) (2022)
- [Differentiable Particle Filtering without Modifying the Forward Pass](https://mlsanity.com/paper/2106.10314/) (2021)
- [Influencing Long-Term Behavior in Multiagent Reinforcement Learning](https://mlsanity.com/paper/2203.03535/) (2022)
- [ROMAX: Certifiably Robust Deep Multiagent Reinforcement Learning via Convex Relaxation](https://mlsanity.com/paper/2109.06795/) (2021)
- [Critic Sequential Monte Carlo](https://mlsanity.com/paper/2205.15460/) (2022)
- [Imperative Learning: A Self-supervised Neuro-Symbolic Learning Framework for Robot Autonomy](https://mlsanity.com/paper/2406.16087/) (2024)
- [A unified view of likelihood ratio and reparameterization gradients](https://mlsanity.com/paper/2105.14900/) (2021)
- [A Theoretical Understanding of Gradient Bias in Meta-Reinforcement Learning](https://mlsanity.com/paper/2112.15400/) (2021)
- [Storchastic: A Framework for General Stochastic Automatic Differentiation](https://mlsanity.com/paper/2104.00428/) (2021)
- [Incentivizing honest performative predictions with proper scoring rules](https://mlsanity.com/paper/2305.17601/) (2023)
- [ULLER: A Unified Language for Learning and Reasoning](https://mlsanity.com/paper/2405.00532/) (2024)
- [Differentiable Simulations for Enhanced Sampling of Rare Events](https://mlsanity.com/paper/2301.03480/) (2023)
- [LOQA: Learning with Opponent Q-Learning Awareness](https://mlsanity.com/paper/2405.01035/) (2024)
- [Advantage Alignment Algorithms](https://mlsanity.com/paper/2406.14662/) (2024)
- [Enabling Realtime Reinforcement Learning at Scale with Staggered Asynchronous Inference](https://mlsanity.com/paper/2412.14355/) (2024)
- [Meta-Value Learning: a General Framework for Learning with Learning Awareness](https://mlsanity.com/paper/2307.08863/) (2023)
- [Unifying Gradient Estimators for Meta-Reinforcement Learning via Off-Policy Evaluation](https://mlsanity.com/paper/2106.13125/) (2021)
- [Biased Gradient Estimate with Drastic Variance Reduction for Meta Reinforcement Learning](https://mlsanity.com/paper/2112.07328/) (2021)
- [An Investigation of the Bias-Variance Tradeoff in Meta-Gradients](https://mlsanity.com/paper/2209.11303/) (2022)
- [Designing Biological Sequences via Meta-Reinforcement Learning and Bayesian Optimization](https://mlsanity.com/paper/2209.06259/) (2022)
- [Probabilistic Programming with Programmable Variational Inference](https://mlsanity.com/paper/2406.15742/) (2024)
- [Multi-agent cooperation through learning-aware policy gradients](https://mlsanity.com/paper/2410.18636/) (2024)

## Abstract (arXiv)

The score function estimator is widely used for estimating gradients of stochastic objectives in stochastic computation graphs (SCG), eg, in reinforcement learning and meta-learning. While deriving the first-order gradient estimators by differentiating a surrogate loss (SL) objective is computationally and conceptually simple, using the same approach for higher-order derivatives is more challenging. Firstly, analytically deriving and implementing such estimators is laborious and not compliant with automatic differentiation. Secondly, repeatedly applying SL to construct new objectives for each order derivative involves increasingly cumbersome graph manipulations. Lastly, to match the first-order gradient under differentiation, SL treats part of the cost as a fixed sample, which we show leads to missing and wrong terms for estimators of higher-order derivatives. To address all these shortcomings in a unified way, we introduce DiCE, which provides a single objective that can be differentiated repeatedly, generating correct estimators of derivatives of any order in SCGs. Unlike SL, DiCE relies on automatic differentiation for performing the requisite graph manipulations. We verify the correctness of DiCE both through a proof and numerical evaluation of the DiCE derivative estimates. We also use DiCE to propose and evaluate a novel approach for multi-agent learning. Our code is available at https://www.github.com/alshedivat/lola.