mirror of
https://github.com/BerriAI/litellm.git
synced 2026-09-16 23:41:43 +00:00
The logging client registers the suite's Claude deployments at build time and
tests call models by their plain provider-qualified names. Phoenix read-back
uses the same env vars as the proxy's arize_phoenix integration and matches
generation spans by span kind plus a unique prompt marker, since otel_v2 names
spans "chat {model}" and leaves successful spans' status UNSET. The compose
stack passes the phoenix env through and opts into span message content so
prompts and completions land on traces; the shared conftest loads tests/e2e/.env
so bare pytest runs work locally
|
||
|---|---|---|
| .. | ||
| conftest.py | ||
| logging_client.py | ||
| test_langfuse_e2e.py | ||
| test_logging_arize_phoenix.py | ||
| test_prometheus_cardinality_e2e.py | ||