A repository of academic publications from the Machine and Human Intelligence Group converted to LLM-friendly text-only Markdown format.
This repository contains research papers converted to plain text with AI-generated descriptions of figures, making them easily accessible for large language model (LLM) analysis and interactions, for both humans and machines.
The full list of papers is available below.
For practical usage, each paper is available in full as well as split into three parts:
| Part | Description | Example |
|---|---|---|
| Main Text | The core content of the paper. | main |
| Backmatter | References, acknowledgments, and other auxiliary content rarely fed to an LLM. | backmatter |
| Appendix | Supplementary materials, when available. | appendix |
| Full Text | Combined version with all parts in a single document. | full |
- Quick usage: Navigate to the paper of interest, click "Copy raw file" on GitHub, paste the full content or individual parts and excerpts into your LLM chat to ask questions about the paper.
- Luigi's usage: Include relevant papers in project repositories for use with advanced LLM assistants. Luigi uses Athanor (an in-house LLM research and coding assistant), but other options include Aider, Cline, Claude Code, and keep growing.
The paper-to-Markdown conversion process uses paper2llm, with Mistral OCR for text and table extraction and Gemini 2.0 Flash for image-to-text descriptions.
Important notes about conversion accuracy.
- Papers have been converted automatically with minimal human intervention.
- OCR models have now become extremely robust, and vision models show practical utility in image understanding, but occasional inaccuracies may occur.
- Errors may take the form of missing sentences near non-standard page formatting, typos in equations or tables, or image descriptions missing or misrepresenting parts of the figure.
- Please report such mistakes by raising a GitHub issue.
For non-critical applications, we consider that the benefit of having LLM-friendly access to research papers outweigh the potential inaccuracies, which generally do not affect the gist of the paper. As usual, double-check key assumptions and results.
-
Amortized Probabilistic Conditioning for Optimization, Simulation and Inference
Chang PE, Loka N, Huang D, Remes U, Kaski S & Acerbi L
AISTATS| main | backmatter | appendix | full -
Normalizing Flow Regression for Bayesian Inference with Offline Likelihood Evaluations
Li C, Huggins B, Mikkola P & Acerbi L
AABI| main | backmatter | appendix | full -
Stacking Variational Bayesian Monte Carlo
Silvestrin F, Li C & Acerbi L
AABI Workshop| main | backmatter | appendix | full -
Inference-Time Prior Adaptation in Simulation-Based Inference via Guided Diffusion Models
Chang PE, Rissanen S, Loka NRBS, Huang D & Acerbi L
AABI Workshop| main | backmatter | appendix | full
-
Improving robustness to corruptions with multiplicative weight perturbations
Trinh T, Heinonen M, Acerbi L & Kaski S
NeurIPS| main | backmatter | appendix | full -
Amortized Bayesian Experimental Design for Decision-Making
Huang D, Guo Y, Acerbi L & Kaski S
NeurIPS| main | backmatter | appendix | full -
Preferential Normalizing Flows
Mikkola P, Acerbi L & Klami A
NeurIPS| main | backmatter | appendix | full -
Amortized Bayesian Workflow (Extended Abstract)
Schmitt M, Li C, Vehtari A, Acerbi L, Burkner P & Radev ST
NeurIPS Workshop| main | backmatter | appendix | full -
Amortized Decision-Aware Bayesian Experimental Design
Huang D, Guo Y, Acerbi L & Kaski S
NeurIPS Workshop| main | backmatter | appendix | full -
Input-gradient space particle inference for neural network ensembles
Trinh T, Heinonen M, Acerbi L & Kaski S
ICLR| main | backmatter | appendix | full -
PyBADS: Fast and robust black-box optimization in Python
Singh G & Acerbi L
JOSS| main | backmatter | full
-
Practical Equivariances via Relational Conditional Neural Processes
Huang D, Hausmann M, Remes U, Clarté G, Luck KS, Kaski S & Acerbi L
NeurIPS| main | backmatter | appendix | full -
Learning Robust Statistics for Simulation-based Inference under Model Misspecification
Huang D, Bharti A, Souza A, Acerbi L & Kaski S
NeurIPS| main | backmatter | appendix | full -
Online Simulator-Based Experimental Design for Cognitive Model Selection
Aushev A, Putkonen A, Clarte G, Chandramouli SH, Acerbi L, Kaski S & Howes A
Comput Brain Behav| main | backmatter | appendix | full -
PyVBMC: Efficient Bayesian inference in Python
Huggins B, Li C, Tobaben M, Aarnos MJ & Acerbi L
JOSS| main | backmatter | full
-
Parallel MCMC Without Embarrassing Failures
de Souza DARMA, Mesquita D, Kaski S & Acerbi L
AISTATS| main | backmatter | appendix | full -
Tackling covariate shift with node-based Bayesian neural networks
Trinh T, Heinonen M, Acerbi L & Kaski S
ICML| main | backmatter | appendix | full
- Uncertainty is maintained and used in working memory
Yoo AH, Acerbi L & Ma WJ
JOV| main | backmatter | full
-
Variational Bayesian Monte Carlo with Noisy Likelihoods
Acerbi L
NeurIPS| main | backmatter | appendix | full -
Dynamic allocation of limited memory resources in reinforcement learning
Patel N, Acerbi L & Pouget A
NeurIPS| main | backmatter | appendix | full -
Unbiased and Efficient Log-Likelihood Estimation with Inverse Binomial Sampling
van Opheusden B, Acerbi L & Ma WJ
PLoS Comput Biol| main | backmatter | appendix | full -
The role of sensory uncertainty in simple contour integration
Zhou Y, Acerbi L & Ma WJ
PLoS Comput Biol| main | backmatter | appendix | full
-
An Exploration of Acquisition and Mean Functions in Variational Bayesian Monte Carlo
Acerbi L
AABI| main | backmatter | appendix | full -
Human online adaptation to changes in prior probability
Norton EH, Acerbi L, Ma WJ & Landy MS
PLoS Comput Biol| main | backmatter | appendix | full
-
Variational Bayesian Monte Carlo
Acerbi L
NeurIPS| main | backmatter | appendix | full -
Bayesian comparison of explicit and implicit causal inference strategies in multisensory heading perception
Acerbi L, Dokka K, Angelaki DE & Ma WJ
PLoS Comput Biol| main | backmatter | appendix | full
-
Practical Bayesian Optimization for Model Fitting with Bayesian Adaptive Direct Search
Acerbi L & Ma WJ
NeurIPS| main | backmatter | appendix | full -
Target Uncertainty Mediates Sensorimotor Error Correction
Acerbi L, Sethu V & Wolpert DM
PLoS ONE| main | backmatter | appendix | full
-
A Framework for Testing Identifiability of Bayesian Models of Perception
Acerbi L, Ma WJ & Vijayakumar S
NeurIPS| main | backmatter | appendix | full -
On the Origins of Suboptimality in Human Probabilistic Inference
Acerbi L, Vijayakumar S & Wolpert DM
PLoS Comput Biol| main | backmatter | appendix | full
- Internal Representations of Temporal Statistics and Feedback Calibrate Motor-Sensory Interval Timing
Acerbi L, Wolpert DM & Vijayakumar S
PLoS Comput Biol| main | backmatter | appendix | full