SOURCE-LINKED INTELLIGENCE

OPEN-1B: A Fully Auditable Training Run

arXiv · AI, language, vision and robotics · article · Sep 15, 2026 · UTC

Open-source language models have a reproducibility problem. Despite releasing weights, training data, and recipes, none of them are provably reproducible due to the non-associativity of floating-point arithmetic. Deep learning frameworks often offer a deterministic execution mode, allowing reproducible operations on the same machines. Unfortunately, this determinism does not carry across hardware such that a user can verify that a released checkpoint was actually produced using the declared training recipe. This leaves room for undisclosed data, injected biases, or backdoors that existing tech

Read original source ↗ Open in workspace

recordType

paper

region

Global