scripts(evals): more eval prompts improvements

changelog/3248.added.md

@@ -1 +1 @@
-- Added pipecat library version info to the `about` field in the `bot-ready` RTVI message
+- Added Pipecat library version info to the `about` field in the `bot-ready` RTVI message.

examples/foundational/07m-interruptible-aws-strands.py

@@ -71,9 +71,9 @@ def build_agent(model_id: str, max_tokens: int):
     @tool
     def check_weather(location: str) -> str:
         if location.lower() == "san francisco":
-            return "The weather in San Francisco is sunny and 30 degrees."
+            return "The weather in San Francisco is sunny and 75 degrees."
         elif location.lower() == "sydney":
-            return "The weather in Sydney is cloudy and 20 degrees."
+            return "The weather in Sydney is cloudy and 60 degrees."
         else:
             return "I'm not sure about the weather in that location."
 

scripts/evals/eval.py

@@ -260,15 +260,21 @@ async def run_eval_pipeline(
 
     eval_function = FunctionSchema(
         name="eval_function",
-        description="Called when the user answers a question.",
+        description=(
+            "Determines whether the user's response satisfies the evaluation "
+            "criteria defined for the current prompt or interaction."
+        ),
         properties={
             "result": {
                 "type": "boolean",
-                "description": "Whether the answer is correct or not",
+                "description": "Whether the user's response meets the evaluation criteria.",
             },
             "reasoning": {
                 "type": "string",
-                "description": "Why the answer was considered correct or invalid",
+                "description": (
+                    "A concise explanation of how the user's response did or did "
+                    "not satisfy the evaluation criteria."
+                ),
             },
         },
         required=["result", "reasoning"],

scripts/evals/run-release-evals.py

@@ -64,13 +64,13 @@ def EVAL_VISION_IMAGE(*, eval_speaks_first: bool = False):
 
 EVAL_VOICEMAIL = EvalConfig(
     prompt="Please leave a message.",
-    eval="The user leaves a voicemail message.",
+    eval="The user provides a reasonable voicemail message.",
     eval_speaks_first=True,
 )
 
 EVAL_CONVERSATION = EvalConfig(
     prompt="Hello, this is Mark.",
-    eval="The user acknowledges the greeting.",
+    eval="The user provides any reasonable conversational response to the greeting.",
     eval_speaks_first=True,
 )
 

src/pipecat/processors/frameworks/rtvi.py

@@ -1423,7 +1423,7 @@ class RTVIProcessor(FrameProcessor):
 
         Args:
             about: Optional information about the bot to include in the ready message.
-                   If left as None, the pipecat library and version will be used.
+                   If left as None, the Pipecat library and version will be used.
         """
         self._bot_ready = True
         # Only call the (deprecated) _update_config method if the we're using a