REVIEW 6 cited by
What does it take to catch a Chinchilla? Verifying Rules on Large-Scale Neural Network Training via Compute Monitoring
Not yet reviewed by Pith; the record is open.
This paper has not been read by Pith yet. Machine review is queued; the pith claim, tier, and objections will appear here once it completes.
SPECIMEN: schema-true, not a live event
T0 review · schema-true
One-sentence machine reading of the paper's core claim.
pith:XXXXXXXX · record.json · timestamp
What does it take to catch a Chinchilla? Verifying Rules on Large-Scale Neural Network Training via Compute Monitoring
read the original abstract
As advanced machine learning systems' capabilities begin to play a significant role in geopolitics and societal order, it may become imperative that (1) governments be able to enforce rules on the development of advanced ML systems within their borders, and (2) countries be able to verify each other's compliance with potential future international agreements on advanced ML development. This work analyzes one mechanism to achieve this, by monitoring the computing hardware used for large-scale NN training. The framework's primary goal is to provide governments high confidence that no actor uses large quantities of specialized ML chips to execute a training run in violation of agreed rules. At the same time, the system does not curtail the use of consumer computing devices, and maintains the privacy and confidentiality of ML practitioners' models, data, and hyperparameters. The system consists of interventions at three stages: (1) using on-chip firmware to occasionally save snapshots of the the neural network weights stored in device memory, in a form that an inspector could later retrieve; (2) saving sufficient information about each training run to prove to inspectors the details of the training run that had resulted in the snapshotted weights; and (3) monitoring the chip supply chain to ensure that no actor can avoid discovery by amassing a large quantity of un-tracked chips. The proposed design decomposes the ML training rule verification problem into a series of narrow technical challenges, including a new variant of the Proof-of-Learning problem [Jia et al. '21].
Forward citations
Cited by 6 Pith papers
-
Macro-Prudential AI Governance: A Two-Layer Early Warning and Response System for Frontier AI
A Basel-III-style two-layer system—coordinated finder-coordinator-defender reporting plus ECAR, CRTH, and ARS buffers—can detect and dampen correlated risk build-up across frontier AI labs’ internal deployments.
-
Detecting Hidden ML Training With Zero-Overhead Telemetry
A classifier using NVML telemetry identifies ML training workloads at 98.2% accuracy and retains 43-87% accuracy against the strongest tested adversarial evasions across 9 GPUs and 5 iteration rounds.
-
How to Catch a GPU: A Taxonomy of Verification and Enforcement Mechanisms for International AI Agreements
Verification of international AI agreements will fail first at detecting hidden compute facilities, around the 10,000-H100-equivalent scale, before other enforcement mechanisms break.
-
Hardware-Level Governance of AI Compute: A Feasibility Taxonomy for Regulatory Compliance and Treaty Verification
Proposes a feasibility taxonomy of 20 hardware-level AI compute governance mechanisms organized by monitoring, verification, and enforcement, with mappings to regulatory scenarios that highlight immaturity of treaty-v...
-
AI Integrity: Defending Against Backdoors and Secret Loyalties
The report defines AI integrity threats (model sabotage and subversion) and recommends four US government policy actions to defend frontier AI systems against backdoors and secret loyalties.
-
An Overview of Catastrophic AI Risks
The paper categorizes sources of catastrophic AI risks into malicious use, AI race, organizational risks, and rogue AIs, providing illustrative stories and mitigation suggestions for each.
discussion (0)
Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.