Fix typing for lower python versions

Author: benomahony

docs/evals.md

@@ -768,13 +768,13 @@ _(This example is complete, it can be run "as is" — you'll need to add `asynci
 
 ### Generating from an Existing Agent
 
-If you already have an agent, you can use [`generate_evals_from_agent`][pydantic_evals.generation.generate_evals_from_agent] to automatically extract types from the agent and generate test cases. This is simpler than `generate_dataset` because you don't need to manually specify the dataset type or generic parameters.
+If you already have an agent, you can use [`generate_dataset_for_agent`][pydantic_evals.generation.generate_dataset_for_agent] to automatically extract types from the agent and generate test cases. This is simpler than `generate_dataset` because you don't need to manually specify the dataset type or generic parameters.
 
 ```python {title="generate_from_agent_example.py"}
 from pydantic import BaseModel
 from pydantic_ai import Agent
 
-from pydantic_evals.generation import generate_evals_from_agent
+from pydantic_evals.generation import generate_dataset_for_agent
 
 
 class AnswerOutput(BaseModel):
@@ -792,7 +792,7 @@ agent = Agent(  # (1)!
 
 
 async def main():
-    dataset = await generate_evals_from_agent(  # (2)!
+    dataset = await generate_dataset_for_agent(  # (2)!
         agent=agent,
         n_examples=3,
         model='openai:gpt-4o',

pydantic_evals/pydantic_evals/generation.py

@@ -10,9 +10,10 @@
 import re
 from collections.abc import Sequence
 from pathlib import Path
-from typing import Any, TypeVar
+from typing import Any
 
 from pydantic import BaseModel, ValidationError
+from typing_extensions import TypeVar
 
 from pydantic_ai import Agent, models
 from pydantic_ai._utils import strip_markdown_fences
@@ -39,6 +40,7 @@ async def generate_dataset(
     extra_instructions: str | None = None,
 ) -> Dataset[InputsT, OutputT, MetadataT]:
     """Use an LLM to generate a dataset of test cases, each consisting of input, expected output, and metadata.
+
     This function creates a properly structured dataset with the specified input, output, and metadata types.
     It uses an LLM to attempt to generate realistic test cases that conform to the types' schemas.