Skip to content

Release 0.50.22

Choose a tag to compare

@github-actions github-actions released this 04 Nov 03:52
· 6 commits to main since this release

Install

Docker

docker pull ghcr.io/jedwards1230/earmark:0.50.22

Helm

helm install earmark oci://ghcr.io/jedwards1230/charts/earmark --version 0.50.22

v0.50.22 Release Notes

This patch release changes the ASR pilot runner (runner/pilot.py) to load models offline from local Hugging Face snapshots, and it lowers the default Granite window length. It also adds step-run tracking for pilot runs and a few new CLI options.

Breaking Changes

  • Granite default window reduced from 240 s to 90 s, and the hard maximum from 540 s to 120 s. Per the runner docs, longer windows caused re-reading and dropped speech. Canary is unchanged (38 s default, 40 s max).
    • Migration: if you set PILOT_WINDOW_SEC for Granite to a value above 120, the run now fails with a config error. Set it to 120 or lower.
    • --bench-max-sec no longer defaults to 1800. It now defaults to max(--bench-batches) × 2 × longest window. Pass it explicitly to keep a fixed cap.
  • Offline model loading requires prefetched snapshots. Runs set HF_HUB_OFFLINE=1 and read model directories from the local HF cache. They no longer download on demand. A missing snapshot or file fails the load with an error naming the asr_pilot_windows_<role>_allow_patterns variable to fix.
    • Migration: prefetch the snapshots before running. Run python pilot.py --backend <granite|canary> --print-prefetch-spec to list the repos, revisions, and allow-patterns that pilot.py needs.
    • Canary also needs the Qwen/Qwen3-1.7B snapshot (revision 70d244cc86ccca08cf5af4e1e306ecf908b1ad5e, or override with PILOT_QWEN_REVISION).
  • --bench behavior changed. It gathers windows across the --asin tracks instead of using only the first track, and it ignores --limit-tracks.

Features

  • Offline model loading. Models load from local snapshot directories in the Hugging Face cache, and --print-prefetch-spec prints the required snapshots as JSON. Canary's LLM config is pointed at the local Qwen snapshot, so no repo lookup happens at load time.
  • Pilot runs appear in run tracking. A real run writes one step_runs row (step asr_pilot), so it shows in the dashboard's Running now panel and in GET /api/v1/runs. It reports progress, and it is closed as done, failed, or cancelled. Dry runs, --bench, and --self-check record nothing.
  • --bench-windows sweeps several window lengths (for example 60,90,120) in one bench run. Bench output now includes per-row degenerate rate and word completeness.
  • --no-repeat-ngram-size N (Granite, default off) is available for experiments.

Improvements

  • Granite stops each batch row on its own once it repeats an 8-token n-gram three times, so one re-reading row no longer holds the whole batch.
  • Degenerate-window detection now also flags re-reads, meaning the same 8-word run three times or the same 16-word run twice. It also flags windows with more than 4 words per audio second as overlong.
  • Special tokens that leak into the output (such as <|fim_middle|>) are stripped from the text of both backends.

Commits

  • chore(earmark): bump chart version to 0.50.22 (f79e1a6)
  • fix(runner): load pilot models from local HF snapshots, offline (Canary self-check) (#223) (342f6ab)

Full Changelog: v0.50.21...v0.50.22