Saved in:
| Main Authors: | , , , , , , , |
|---|---|
| Format: | Preprint |
| Published: |
2026
|
| Subjects: | |
| Online Access: | https://arxiv.org/abs/2602.12630 |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| _version_ | 1866910021093163008 |
|---|---|
| author | Baser, Oguzhan Sadeghi, Elahe Wang, Eric Alves, David Ribeiro Kazemian, Sam Kang, Hong Chinchali, Sandeep P. Vishwanath, Sriram |
| author_facet | Baser, Oguzhan Sadeghi, Elahe Wang, Eric Alves, David Ribeiro Kazemian, Sam Kang, Hong Chinchali, Sandeep P. Vishwanath, Sriram |
| contents | Most large language models (LLMs) run on external clouds: users send a prompt, pay for inference, and must trust that the remote GPU executes the LLM without any adversarial tampering. We critically ask how to achieve verifiable LLM inference, where a prover (the service) must convince a verifier (the client) that an inference was run correctly without rerunning the LLM. Existing cryptographic works are too slow at the LLM scale, while non-cryptographic ones require a strong verifier GPU. We propose TensorCommitments (TCs), a tensor-native proof-of-inference scheme. TC binds the LLM inference to a commitment, an irreversible tag that breaks under tampering, organized in our multivariate Terkle Trees. For LLaMA2, TC adds only 0.97% prover and 0.12% verifier time over inference while improving robustness to tailored LLM attacks by up to 48% over the best prior work requiring a verifier GPU. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2602_12630 |
| institution | arXiv |
| publishDate | 2026 |
| record_format | arxiv |
| spellingShingle | TensorCommitments: A Lightweight Verifiable Inference for Language Models Baser, Oguzhan Sadeghi, Elahe Wang, Eric Alves, David Ribeiro Kazemian, Sam Kang, Hong Chinchali, Sandeep P. Vishwanath, Sriram Cryptography and Security Artificial Intelligence Most large language models (LLMs) run on external clouds: users send a prompt, pay for inference, and must trust that the remote GPU executes the LLM without any adversarial tampering. We critically ask how to achieve verifiable LLM inference, where a prover (the service) must convince a verifier (the client) that an inference was run correctly without rerunning the LLM. Existing cryptographic works are too slow at the LLM scale, while non-cryptographic ones require a strong verifier GPU. We propose TensorCommitments (TCs), a tensor-native proof-of-inference scheme. TC binds the LLM inference to a commitment, an irreversible tag that breaks under tampering, organized in our multivariate Terkle Trees. For LLaMA2, TC adds only 0.97% prover and 0.12% verifier time over inference while improving robustness to tailored LLM attacks by up to 48% over the best prior work requiring a verifier GPU. |
| title | TensorCommitments: A Lightweight Verifiable Inference for Language Models |
| topic | Cryptography and Security Artificial Intelligence |
| url | https://arxiv.org/abs/2602.12630 |