wx44wx/pipecat
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

Update evals to include examples that exercise the universal LLMContext

Browse Source
This commit is contained in:
Paul Kompfner
2025-09-11 14:32:56 -04:00
parent 9b3f6927c2
commit a4023d3915
1 changed files with 19 additions and 0 deletions
Download Patch File Download Diff File Expand all files Collapse all files

19
scripts/evals/run-release-evals.py
View File

@@ -135,6 +135,25 @@ TESTS_14 = [
("14r-function-calling-aws.py", PROMPT_WEATHER, EVAL_WEATHER, BOT_SPEAKS_FIRST),
("14v-function-calling-openai.py", PROMPT_WEATHER, EVAL_WEATHER, BOT_SPEAKS_FIRST),
("14w-function-calling-mistral.py", PROMPT_WEATHER, EVAL_WEATHER, BOT_SPEAKS_FIRST),
("14x-function-calling-universal-context.py", PROMPT_WEATHER, EVAL_WEATHER, BOT_SPEAKS_FIRST),
(
"14y-function-calling-google-universal-context.py",
PROMPT_WEATHER,
EVAL_WEATHER,
BOT_SPEAKS_FIRST,
),
(
"14z-function-calling-anthropic-universal-context.py",
PROMPT_WEATHER,
EVAL_WEATHER,
BOT_SPEAKS_FIRST,
),
(
"14aa-function-calling-aws-universal-context.py",
PROMPT_WEATHER,
EVAL_WEATHER,
BOT_SPEAKS_FIRST,
),
# Currently not working.
# ("14c-function-calling-together.py", PROMPT_WEATHER, EVAL_WEATHER, BOT_SPEAKS_FIRST),
# ("14l-function-calling-deepseek.py", PROMPT_WEATHER, EVAL_WEATHER, BOT_SPEAKS_FIRST),