troysaved/claimtrace-smoke
troysaved/claimtrace-smoke is a 2 billion parameter language model developed by troysaved, primarily serving as a throwaway checkpoint for testing publishing pipelines. This model was trained on a small smoke batch, not the full dataset, and does not exhibit meaningful behavior or represent the final claimtrace model. It is maintained publicly to prevent broken links from early testing phases.
Loading preview...
Overview
This model, troysaved/claimtrace-smoke, is explicitly not the intended claimtrace submission. It is a 2 billion parameter checkpoint created by troysaved solely for smoke-testing a publishing pipeline on 2026-08-19. It was trained on a minimal batch of data, not the actual dataset, and therefore lacks any meaningful or intended behavior.
Key Characteristics
- Purpose: A temporary artifact to confirm the publishing process for the
claimtraceproject. - Training: Trained on a small "smoke batch" rather than the full
claimtrace-ledger-dataset. - Behavior: Does not possess any significant or intended functional behavior.
- Status: Kept public to ensure that any links generated during the testing phase resolve to an explanation rather than a 404 error.
Important Note
Users looking for the actual claimtrace model should refer to troysaved/claimtrace-qwen3-1.7b. This smoke version should not be evaluated or used for any practical applications.