mlsanityData to 14 Oct 2026

TAO: Tolerance-Aware Optimistic Verification for Floating-Point Neural Networks

arXiv 2510.16028 ↗First posted 15 Oct 2025cs.CR cs.AI cs.LGcited by 1

Not on the map. Why?

Abstract (arXiv)

Neural networks increasingly run on hardware outside the user's control (cloud GPUs, inference marketplaces). Yet ML-as-a-Service reveals little about what actually ran or whether returned outputs faithfully reflect the intended inputs. Users lack recourse against service downgrades (model swaps, quantization, graph rewrites, or discrepancies like altered ad embeddings). Verifying outputs is hard because floating-point(FP) execution on heterogeneous accelerators is inherently nondeterministic. Existing approaches are either impractical for real FP neural networks or reintroduce vendor trust. We present TAO: a Tolerance Aware Optimistic verification protocol that accepts outputs within principled operator-level acceptance regions rather than requiring bitwise equality. TAO combines two error models: (i) sound per-operator IEEE-754 worst-case bounds and (ii) tight empirical percentile profiles calibrated across hardware. Discrepancies trigger a Merkle-anchored, threshold-guided dispute game that recursively partitions the computation graph until one operator remains, where adjudication reduces to a lightweight theoretical-bound check or a small honest-majority vote against empirical thresholds. Unchallenged results finalize after a challenge window, without requiring trusted hardware or deterministic kernels. We implement TAO as a PyTorch-compatible runtime and a contract layer currently deployed on Ethereum Holesky testnet. The runtime instruments graphs, computes per-operator bounds, and runs unmodified vendor kernels in FP32 with negligible overhead (0.3% on Qwen3-8B). Across CNNs, Transformers and diffusion models on A100, H100, RTX6000, RTX4090, empirical thresholds are 102-103 times tighter than theoretical bounds, and bound-aware adversarial attacks achieve 0% success. Together, TAO reconciles scalability with verifiability for real-world heterogeneous ML compute.

Key facts. Cited by 1 paper in the map (data through 14 Oct 2026). Builds on 25 of its 62 references that are in the map. First posted on arXiv 15 Oct 2025.

Builds on

25 of 62 references.

References · 25

  1. 2015Deep Residual Learning for Image Recognitioncited by 52,096
  2. 2018BERT: Pre-training of Deep Bidirectional Transformers for Language Understandingcited by 40,399
  3. 2019Exploring the Limits of Transfer Learning with a Unified Text-to-Text Transformercited by 14,599
  4. 2016Pointer Sentinel Mixture Modelscited by 3,096
  5. 2023Atom: Low-bit Quantization for Efficient and Accurate LLM Servingcited by 272
  6. 2018Slalom: Fast, Verifiable and Private Execution of Neural Networks in Trusted Hardwarecited by 102
  7. 2021Proof-of-Learning: Definitions and Practicecited by 84
  8. 2017SafetyNets: Verifiable Execution of Deep Neural Networks on an Untrusted Cloudcited by 50
  9. 2020Exploiting Verified Neural Networks via Floating Point Numerical Errorcited by 35
  10. 2018I Know What You See: Power Side-Channel Attack on Convolutional Neural Network Acceleratorscited by 27
  11. 2023Tools for Verifying Neural Models' Training Datacited by 22
  12. 2024opML: Optimistic Machine Learning on Blockchaincited by 21
13 more
  1. 2019A scalable verification solution for blockchainscited by 10
  2. 2022ezDPS: An Efficient and Zero-Knowledge Machine Learning Inference Pipelinecited by 10
  3. 20213LegRace: Privacy-Preserving DNN Training over TEEs and GPUscited by 9
  4. 2025Verde: Verification via Refereed Delegation for Machine Learning Programscited by 9
  5. 2024Optimistic Verifiable Training by Controlling Hardware Nondeterminismcited by 7
  6. 2021Agatha: Smart Contract for DNN Computationcited by 6
  7. 2025No Soundness in the Real World: On the Challenges of the Verification of Deployed Neural Networkscited by 6
  8. 2024Proof-of-Learning with Incentive Securitycited by 4
  9. 2024Revealing Floating-Point Accumulation Orders in Software/Hardware Implementationscited by 4
  10. 2025PTEENet: Post-Trained Early-Exit Neural Networks Augmentation for Inference Cost Optimizationcited by 2
  11. 2024Proof of Diligence: Cryptoeconomic Security for Rollupscited by 1
  12. 2024SERENE: A Collusion Resilient Replication-based Verification Frameworkcited by 1
  13. 2024Proof of Sampling: A Nash Equilibrium-Based Verification Protocol for Decentralized Systemscited by 1

Built on by

Cited by year (2026 to 14 Oct)

Citing papers per year: 2026: 1.

1 paper.

  1. 2026IMMACULATE: A Practical LLM Auditing Framework via Verifiable Computationcited by 3

Similar, not cited

Close in wording, no citation either way

  1. 2026Lipschitz-Based Robustness Certification Under Floating-Point Execution
  2. 2026From Bit-Position Sensitivity to Unequal Error Protection for DNN Inference Memory
  3. 2022QVIP: An ILP-based Formal Verification Approach for Quantized Neural Networks
  4. 2026Deterministic LLM Inference Across GPU Kernels: Power-of-Two INT8 Quantization Scales and the Limits of Tolerance-Based Conformance
  5. 2025Bridging Efficiency and Safety: Formal Verification of Neural Networks with Early Exits
  6. 2021Verifying Quantized Neural Networks using SMT-Based Model Checking
  7. 2024Verifying Properties of Binary Neural Networks Using Sparse Polynomial Optimization
  8. 2022Adaptive Block Floating-Point for Analog Deep Learning Hardware