An easy-to-use Python library for processing and manipulating 3D point clouds and meshes.

Optimal transport tools implemented with the JAX framework, to solve large scale matching problems of any flavor.

FlashSinkhorn: IO-Aware Entropic Optimal Transport in PyTorch + Triton. Streaming Sinkhorn with O(nd) memory.

FML (Francis' Machine-Learnin' Library) - A collection of utilities for machine learning tasks

Interactive visualization of Manifold-Constrained Hyper-Connections (mHC) for stable deep network training

A thorough review of the paper "Learning Embeddings into Entropic Wasserstein Spaces" by Frogner et al. Includes a reproduction of the results on word embeddings.

Tensorflow implementation of optimal transport (OT) with Sinkhorn algorithm.

Code for ECML/PKDD paper: "LSMI-Sinkhorn: Semi-supervised Mutual Information Estimation with Optimal Transport"

Douglas-Rachford Splitting for Optimal Transport

No description provided.

Optimal Transport and Optimization related experiments.

Code for the paper Geodesic Sinkhorn for Fast and Accurate Optimal Transport on Manifolds.

JAX implementation of DETR

Implementation of the paper "Anticipation-Free Training for Simultaneous Machine Translation"

A dedicated convenient repo for different Music Transformers implementations (Reformer/XTransformer/Sinkhorn/etc)

No description provided.

[ICML 2023] Official code for our paper: 'Conditional Tree Matching for Inference-Time Adaptation of Tree Prediction Models'

Code for "A MFG Model for the Dynamics of Cities" (Barilla, Carlier, Lasry 2021)

Library for solving variational mean-field games using optimal transport and the Sinkhorn algorithm

MLX + Metal implementation of mHC: Manifold-Constrained Hyper-Connections by DeepSeek-AI.

Comparing ResNet vs Hyper-Connections vs causal Sinkhorn mHC on Fashion-MNIST using MLX.

A neural parser for typelogical grammars based on Sinkhorn networks and Linear Logic Proof Nets.

Code for some of the algorithms used in Optimal Transport like subspace alignment and sinkhorn algorithms

Quantum-Entropic Martingale Transport for Robust Model-Free Derivative Pricing.

This project enhances text encoders' global descriptors by implementing advanced aggregation techniques from computer vision literature, to improve sentence-level representations. We utilize models such as BERT, RoBERTa, and CLIP, and benchmark performance using datasets like MTEB, SemEval24 and Quora Question Pairs.

🔍 Explore mHC for manifold-constrained hyper-connections in PyTorch, enhancing deep learning layer updates with efficient, non-negative mixing maps.

High-performance Triton kernels for balanced and unbalanced Sinkhorn optimal transport.