mlsanityData to 7 Oct 2026

GenomeOcean Anywhere: Private WebGPU Inference for Genome MoEs

arXiv 2609.35882 ↗First posted 27 Sep 2026cs.CR cs.LG q-bio.GNcited by 0

On the map: Mixture-of-experts → Efficient MoE training and inference systems

Abstract (arXiv)

Genome foundation models are most useful where sequences are generated, yet the largest models need datacenter accelerators and a place to send private DNA. We ask whether a 15-billion-parameter genome mixture-of-experts (MoE) model can instead run on volunteers' web browsers, with the experts spread across many untrusted devices, without changing its predictions and without revealing the sequence to any single device. We build a system in which a trusted coordinator runs attention and routing while browser workers run every expert feed-forward network through hand-written WebGPU kernels, and we protect the expert inputs with real-valued Lagrange coded computing: each worker receives only a Gaussian-padded share, computes the expert's linear maps, and the coordinator decodes from any two of three workers. On GenomeOcean-MoE (8 experts, top-2 routing, 24 layers), the browser path matches native llama.cpp at every quantization level, the distributed path stays at the BF16 numerical noise floor (KL 0.0036 nats per token), and an unfitted latency model predicts decode time within 0.74% (median) under emulated wide-area links. We first show that plaintext expert inputs are not private: a probe recovers the token from a single vector at every depth, and one worker can identify the source genome from 300 unordered tokens with 92% accuracy. With coded experts, an adaptive attacker trained on shares falls to the most-frequent-token baseline, one worker's information about each token is bounded below one bit per forward pass, and the fidelity cost stays below the BF16 noise floor; in Chrome, coded decoding runs at 220 to 376 ms per token, depending on how much of the routing is hidden, and continues without replicas when a worker fails.

Key facts. No citing papers in the map yet (data through 7 Oct 2026). Placed in Efficient MoE training and inference systems, one of 12 directions in Mixture-of-experts. Builds on 41 of its 57 references that are in the map. First posted on arXiv 27 Sep 2026.

Builds on

41 of 57 references.

Closest ancestors · 6

Same direction, newest first

  1. 2024WDMoE: Wireless Distributed Large Language Models with Mixture of Experts6 May
  2. 2024Fiddler: CPU-GPU Orchestration for Fast Inference of Mixture-of-Experts Models10 Feb
  3. 2023EdgeMoE: Empowering Sparse Large Language Models on Mobile Devices28 Aug
  4. 2022MegaBlocks: Efficient Sparse Training with Mixture-of-Experts29 Nov
  5. 2022DeepSpeed-MoE: Advancing Mixture-of-Experts Inference and Training to Power Next-Generation AI Scale14 Jan
  6. 2020Towards Crowdsourced Training of Large Neural Networks using Decentralized Mixture-of-Experts10 Feb

Other references · 35

  1. 2017Outrageously Large Neural Networks: The Sparsely-Gated Mixture-of-Experts Layercited by 3,413
  2. 2021Switch Transformers: Scaling to Trillion Parameter Models with Simple and Efficient Sparsitycited by 3,014
  3. 2024Mixtral of Expertscited by 1,785
  4. 2020GShard: Scaling Giant Models with Conditional Computation and Automatic Shardingcited by 1,601
  5. 2018Split learning for health: Distributed deep learning without sharing raw patient datacited by 391
  6. 2022Sparse Upcycling: Training Mixture-of-Experts from Dense Checkpointscited by 207
  7. 2023Text Embeddings Reveal (Almost) As Much As Textcited by 189
  8. 2023HyenaDNA: Long-Range Genomic Sequence Modeling at Single Nucleotide Resolutioncited by 187
  9. 2023DNABERT-2: Efficient Foundation Model and Benchmark For Multi-Species Genomecited by 139
  10. 2021CrypTen: Secure Multi-Party Computation Meets Machine Learningcited by 131
  11. 2024Caduceus: Bi-Directional Equivariant Long-Range DNA Sequence Modelingcited by 108
  12. 2020Unleashing the Tiger: Inference Attacks on Split Learningcited by 104
23 more
  1. 2018Slalom: Fast, Verifiable and Private Execution of Neural Networks in Trusted Hardwarecited by 101
  2. 2022Decentralized Training of Foundation Models in Heterogeneous Environmentscited by 87
  3. 2022Petals: Collaborative Inference and Fine-tuning of Large Modelscited by 84
  4. 2015Speeding Up Distributed Machine Learning Using Codescited by 76
  5. 2022MPCFormer: fast, performant and private Transformer inference with MPCcited by 67
  6. 2018Lagrange Coded Computing: Optimal Design for Resiliency, Security and Privacycited by 66
  7. 2021Distributed Deep Learning in Open Collaborationscited by 58
  8. 2023SWARM Parallelism: Training Large Models Can Be Surprisingly Communication-Efficientcited by 58
  9. 2023Distributed Inference and Fine-tuning of Large Language Models Over The Internetcited by 42
  10. 2017Polynomial Codes: an Optimal Design for High-Dimensional Coded Matrix Multiplicationcited by 32
  11. 2024Galaxy: A Resource-Efficient Collaborative Edge AI System for In-situ Transformer Inferencecited by 25
  12. 2019CodedPrivateML: A Fast and Privacy-Preserving Framework for Distributed Machine Learningcited by 20
  13. 2024Genomic Language Models: Opportunities and Challengescited by 20
  14. 2020Berrut Approximated Coded Computing: Straggler Resistance Beyond Polynomial Computingcited by 19
  15. 2019Moving Deep Learning into Web Browser: How Far Can We Go?cited by 14
  16. 2025Prompt Inversion Attack against Collaborative Inference of Large Language Modelscited by 14
  17. 2025Depth Gives a False Sense of Privacy: LLM Internal States Inversioncited by 14
  18. 2020Analog Lagrange Coded Computingcited by 13
  19. 2024WebLLM: A High-Performance In-Browser LLM Inference Enginecited by 13
  20. 2019BOINC: A Platform for Volunteer Computingcited by 6
  21. 2021ApproxIFER: A Model-Agnostic Approach to Resilient and Robust Prediction Serving Systemscited by 5
  22. 2025CryptoMoE: Privacy-Preserving and Scalable Mixture of Experts Inference via Balanced Expert Routingcited by 4
  23. 2023Empowering In-Browser Deep Learning Inference on Edge Devices with Just-in-Time Kernel Optimizationscited by 3

Built on by

Not cited yet.

Similar, not cited

Close in wording, no citation either way

  1. 2026CipherGenome: Homomorphic Inference for Genomic Mixture-of-Experts
  2. 2026Relay Buffer Independent Communication over Pooled HBM for Efficient MoE Inference on Ascend
  3. 2025GRACE-MoE: Grouping and Replication with Locality-Aware Routing for Efficient Distributed MoE Inference
  4. 2026EasyBalance: Cross-Layer Load Balancing in Distributed MoE Inference
  5. 2024MoE++: Accelerating Mixture-of-Experts Methods with Zero-Computation Experts
  6. 2024Communication-Efficient Sparsely-Activated Model Training via Sequence Migration and Token Condensation
  7. 2026Fast MoE Inference via Predictive Prefetching and Expert Replication
  8. 2025Accelerating Edge Inference for Distributed MoE Models with Latency-Optimized Expert Placement