AI/ML / Developer Tool
An open-source simulator that studies how GPT inference behaves under space radiation and cosmic-ray fault conditions, with CLI runs, comparisons, and self-contained HTML reports.
The page clearly describes a real AI-adjacent developer tool: a GPT-in-space simulator for testing radiation-induced faults in model weights, activations, and KV cache. The repository shows concrete usage examples, supported fault types and environments, report generation, and a visible project structure with tests and docs, which is enough for a useful public listing.
The crawled page does not show a finished product UI or a live demo beyond the linked reports, and it is still focused on a seeded tiny GPT skeleton rather than a larger model backend. It is also not obvious from the snapshot who would use it in practice beyond experimentation or research.
Last evaluated 3 days ago. Current rank #638. Down 1 spot in the rankings.
An open-source project for making AI research agents legible, auditable, and traceable through behavioral code and event logs.
