Post two CI badges on a PR: CPU PR CI and GPU run-slow (#48190)
A PR accumulates two independent CI streams -- regular PR CI on CPU, and the GPU
runs a maintainer asks for with a `run-slow: <models>` comment. A single badge
had to pick one, and picked whichever ran last, so a PR could carry a red badge
sourced from a run-slow run beside a dashboard link explaining a different run.
PR #48171 read "7 errors" against a dashboard reporting none.
Emit one badge per stream, side by side on one line, each asking the exporter
for its own stream via the new ?event= parameter.
Both are always emitted. This workflow only fires on PR CI completion, so a GPU
badge written only when a run-slow run already existed would stay missing after
a run-slow that no push follows -- exactly the #48171 case. The badge SVGs are
live, so the URLs written here pick up new runs of their stream with no rewrite,
and the exporter renders the GPU badge as "not run" until there is something to
report.
Requires the ?event= support in huggingface/transformers-ci; without it both
URLs fall back to the same unfiltered badge, which is today's behaviour.
transformers
e453228e - Post two CI badges on a PR: CPU PR CI and GPU run-slow (#48190)
