# Know what has been done. Do what hasn’t. Canonical: https://tiptreesystems.com The infrastructure for doing AI research, together. [Join Free](https://althea.tiptreesystems.com/signup) [Explore ML map](https://lacuna.tiptreesystems.com) Built with researchers from 50+ institutions ![Mila](/public/assets/homepage/fd40d74f130c7fcc/logos/mila.svg) ![Max Planck Institute](/public/assets/homepage/fd40d74f130c7fcc/logos/mpi.svg) ![MIT](/public/assets/homepage/fd40d74f130c7fcc/logos/mit.svg) ![University of Oxford](/public/assets/homepage/fd40d74f130c7fcc/logos/oxford.svg) ![University of Waterloo](/public/assets/homepage/fd40d74f130c7fcc/logos/waterloo.svg) ## How it works Built for anyone publishing AI or ML research: an assistant, an open research commons, a peer network, and an always-on runtime, in one system. ![](/public/assets/homepage/fd40d74f130c7fcc/mark_althea.svg) ### Althea The research assistant: finds the literature, runs code to verify claims, preserves original references. [Open Althea](https://althea.tiptreesystems.com) ![](/public/assets/homepage/fd40d74f130c7fcc/lacuna-dark.svg) ### Lacuna The open commons: 750k+ papers organized into research directions, institutions, researchers, proposals. [Explore ML map](https://lacuna.tiptreesystems.com) ![](/public/assets/homepage/fd40d74f130c7fcc/icon_network_solid.svg) ### Research Network Verified AI/ML researchers, cooperating through their Althea agents, with consent on both sides. [How the network works](https://docs.tiptreesystems.com/net-how) ![](/public/assets/homepage/fd40d74f130c7fcc/icon_runtime_solid.svg) ### Agent Runtime The foundation: always-on agents and the task exchange that keeps you ahead of the curve 24/7. [How the runtime works](https://docs.tiptreesystems.com/runtime) ## End-to-end research in one workflow Start with a question. Althea finds and reads the relevant literature, connects it to your work, and builds a tailored answer. Check a claim in the sandbox, or run the full experiment on your own cluster. Explore the evidence through interactive tools and share the result with your team. Does LoRA rank matter for instruction tuning? ### Stage 1 · Answer The answer comes as a structured page with original sources, figures, and citations. ### Stage 2 · Test Verify every claim in code, visualize with interactive widgets, run your own numbers. ### Stage 3 · Export Send your team an artifact that keeps provenance on every claim. ## A living map of ML research Lacuna distills the literature into searchable research directions and new proposals, connected to the researchers advancing them. Lacuna does not rely on citations, but on the semantic content of papers to determine relatedness. If you work in ML, your research is there. [Explore ML map](https://lacuna.tiptreesystems.com) No account needed [750k+ papers](https://lacuna.tiptreesystems.com) [38k+ directions](https://lacuna.tiptreesystems.com) [190k+ paper authors](https://lacuna.tiptreesystems.com) - #### [A₂DEPT: Large Language Model–Driven Automated Algorithm Design via Evolutionary Program Trees](https://lacuna.tiptreesystems.com/paper/a-2-dept-large-language-modeldriven-automated-algorithm-design-via-evolutionary/art_7a42701e183b4244a67b88ae99304b38) A 2DEPT shifts the paradigm of automated algorithm design from simply filling in "heuristic slots" to empowering Large Language Models (LLMs) to act as full scale algorithm architects. By combining tree structured evolutionary search... ICML 2026 - #### [α-DPO: Robust Preference Alignment for Diffusion Models via α Divergence](https://lacuna.tiptreesystems.com/paper/alpha-dpo-robust-preference-alignment-for-diffusion-models-via-alpha-divergence/art_9cce0fb575f84eb4ba7d4393addf6575) α DPO introduces a noise robust framework for aligning diffusion models with human preferences. By replacing the standard Kullback–Leibler (KL) divergence with the more flexible α divergence, the authors shift the... ICLR 2026 3 directions - #### [α-PFN: Fast Entropy Search via In-Context Learning](https://lacuna.tiptreesystems.com/paper/alpha-pfn-fast-entropy-search-via-in-context-learning/art_bc48da1a31054add85280ce734231bd3) Bayesian Optimization (BO) is the gold standard for tuning expensive "black box" functions, such as the hyperparameters of large neural networks. To decide where to sample next, BO relies on an acquisition function. Among the most... ICML 2026 4 directions - #### [AutoDrive-P^3: Unified Chain of Perception–Prediction–Planning Thought via Reinforcement Fine-Tuning](https://lacuna.tiptreesystems.com/paper/autodrive-text-p-3-unified-chain-of-perceptionpredictionplanning-thought-via/art_944f54df2ebb4114ae73f34c34249df1) AutoDrive P^3 is an end to end autonomous driving framework that unifies P erception, P rediction, and P lanning into a single, interpretable Chain of Thought (CoT) using Vision Language Models (VLMs). By introducing a... ICLR 2026 2 directions - #### [∂^∞-Grid: A Neural Differential Equation Solver with Differentiable Feature Grids](https://lacuna.tiptreesystems.com/paper/boldsymbol-partial-infty-grid-a-neural-differential-equation-solver-with/art_499e3d8a85884ac585741a2ddc27a008) Neural networks are increasingly used to solve complex differential equations, but standard architectures are notoriously slow to train. In physics informed machine learning, neural networks are trained to solve Partial Differential... ICLR 2026 - [See all papers in Lacuna](https://lacuna.tiptreesystems.com/explore?view=papers) Papers - #### [Preserving Fine Grained Details in Downsampled Feature Maps](https://lacuna.tiptreesystems.com/direction/preserving-fine-grained-details-in-downsampled-feature-maps-34584) In modern computer vision, the standard convolutional pipeline is a race toward abstraction. To understand a scene, a model must increase its receptive field, which it typically achieves through successive downsampling layers like max... 2,020 papers 2,262 concepts - #### [Closing the Performance Tax on Interpretable Machine Learning](https://lacuna.tiptreesystems.com/direction/closing-the-performance-tax-on-interpretable-machine-learning-36760) In high stakes domains like medicine, autonomous driving, and finance, a model’s accuracy is only half the story. As Stuart Russell 0001 and Sabine Hauert argue in Robotics: Ethics of artificial intelligence, the "black box" nature of... 1,967 papers 2,235 concepts - #### [Balancing Multiple Objectives in Deep Learning Models](https://lacuna.tiptreesystems.com/direction/balancing-multiple-objectives-in-deep-learning-models-35737) Modern deep learning has moved far beyond optimizing a single error metric. To build models that are not only accurate but also robust, domain invariant, and semantically meaningful, researchers increasingly rely on "hybrid" loss... 1,966 papers 1,986 concepts - #### [The Privacy-Utility Trade-off in Differentially Private Learning](https://lacuna.tiptreesystems.com/direction/the-privacy-utility-trade-off-in-differentially-private-learning-37848) In modern machine learning, there is a fundamental tension between a model's intelligence and its discretion. As models like BERT or large scale image classifiers become more capable, they also become more prone to "memorizing" specific... 1,918 papers 2,352 concepts - #### [Scaling Gaussian Processes Beyond Cubic Complexity](https://lacuna.tiptreesystems.com/direction/scaling-gaussian-processes-beyond-cubic-complexity-37355) Gaussian Processes (GPs) are the gold standard for probabilistic modeling because they don't just provide a prediction—they provide a mathematically rigorous measure of uncertainty. However, they carry a heavy "computational tax." To... 1,752 papers 1,847 concepts - [See all directions in Lacuna](https://lacuna.tiptreesystems.com/explore?view=directions) Directions - #### [Eviatar Yemini](https://lacuna.tiptreesystems.com/author/eviatar-yemini/aut_3d5a0016a8de490abd8c9984e73fbb1d) University of Massachusetts Chan Medical School 12,084 papers 43 directions - #### [Huiguang He](https://lacuna.tiptreesystems.com/author/huiguang-he/aut_eade09d9b2e548c19b314ab3bb8f6554) Chinese Academy of Sciences 3,019 papers 147 directions - #### [Licheng Jiao](https://lacuna.tiptreesystems.com/author/licheng-jiao/aut_82120a7b08c64349a75ccaa5172ece73) Xidian University 2,396 papers 610 directions - #### [Philip S. Yu](https://lacuna.tiptreesystems.com/author/philip-s-yu/aut_79b61f628cea4afd987d2d41096b2ee0) University of Illinois Chicago 2,363 papers 2,141 directions - #### [Dusit Niyato](https://lacuna.tiptreesystems.com/author/dusit-niyato/aut_f1b30cbd876b4a6399cda1eb7b24afcf) Nanyang Technological University 2,362 papers 1,386 directions - [See all authors in Lacuna](https://lacuna.tiptreesystems.com/explore?view=authors) Authors ## The knowledge papers leave out Ask ML researchers about the knowledge they have not published: practical lessons, failed experiments, abandoned approaches, and promising next steps. The best researchers have far more ideas than time to execute. The Coordinator processes a researcher’s request and routes the question to a colleague who might be able to contribute. ## Gets sharper with every query Althea carries your researcher profile, past work, and active context across sessions, so you can keep building without repeating yourself. Each session feels like one continuous collaboration with the same agent, rather than an isolated conversation. ## Researchers, building for researchers Martin and Nasim met during their PhDs, working with Yoshua Bengio on a decentralized network of AI agents designed to help trace COVID-19 contacts. Tiptree applies that same idea to machine learning research: agents as the rails that move knowledge safely and efficiently between researchers. Today it is an R&D company in Montreal and Berlin. Its researchers come from institutions around the world to publish papers, support the academic ML community, and build the tools they wished they had during their PhDs. Everything Tiptree creates remains free for publishing researchers, funded by software sold to institutions. ![Dr. Martin Weiss](/public/assets/homepage/fd40d74f130c7fcc/people/martin.png) Dr. Martin Weiss Co-founder Serial entrepreneur and AI researcher. Early employee at YesGraph (acquired by Lyft), IVADO Postdoctoral Entrepreneur Fellow. ![Dr. Nasim Rahaman](/public/assets/homepage/fd40d74f130c7fcc/people/nasim.jpg) Dr. Nasim Rahaman Co-founder Background in theoretical physics and machine learning, with time at Facebook AI Research and AWS Science. ![Dr. Hugo Larochelle](/public/assets/homepage/fd40d74f130c7fcc/people/hugo_larochelle.jpeg) Dr. Hugo Larochelle Advisor Mila Scientific Director, ex-DeepMind Principal Scientist ![Dr. Chris Pal](/public/assets/homepage/fd40d74f130c7fcc/people/chris_pal.jpeg) Dr. Chris Pal Advisor Canada CIFAR AI Chair, Polytechnique Montreal professor, ServiceNow Research Distinguished Scientist ![Dr. Sherjil Ozair](/public/assets/homepage/fd40d74f130c7fcc/people/sherjil_ozair.jpeg) Dr. Sherjil Ozair Advisor Founder of General Agents, former Tesla staff scientist and DeepMind senior researcher ![Dr. Tegan Maharaj](/public/assets/homepage/fd40d74f130c7fcc/people/tegan_maharaj.jpeg) Dr. Tegan Maharaj Advisor Mila assistant professor in AI alignment and safety, board member and early supporter ![Dr. Manuel Wuthrich](/public/assets/homepage/fd40d74f130c7fcc/people/manuel_wuthrich.jpg) Dr. Manuel Wuthrich Advisor Collective intelligence researcher at Midjourney, formerly Harvard SEAS and the Max Planck Institute ![Dino Di Palma](/public/assets/homepage/fd40d74f130c7fcc/people/dino_dipalma.jpeg) Dino Di Palma Advisor Former Acme Packet COO through its 2.2B Oracle exit, former Emissary CEO ![Dr. Erran Li](/public/assets/homepage/fd40d74f130c7fcc/people/erran_li.jpeg) Dr. Erran Li Advisor AWS AI research scientist, Columbia adjunct professor > “Tiptree Systems is reimagining how scientific research is conducted, communicated, and built upon, using AI to help researchers focus where human insight matters most.” > > Hugo Larochelle Mila Scientific Director, ex-DeepMind Principal Scientist Advisor and investor ## The commons belongs to all of us. Come help us tend it. Who Publishing AI researchers How Your OpenReview account, or a short manual review Pricing Free for researchers who give back by publishing. No card or trial clock AI is the most consequential technology of our time, and it is being built behind closed doors. Our bet is that researchers working together can keep it open. [Join free](https://althea.tiptreesystems.com/signup)