Fix OpenAI completion args routing

evals/cli/oaieval.py

@@ -22,6 +22,23 @@ def _purple(str: str) -> str:
     return f"\033[1;35m{str}\033[0m"
 
 
+def parse_key_value_args(param_str: str) -> dict[str, Any]:
+    def to_number(x: str) -> Union[str, int, float]:
+        try:
+            return int(x)
+        except (ValueError, TypeError):
+            pass
+
+        try:
+            return float(x)
+        except (ValueError, TypeError):
+            pass
+        return x
+
+    str_dict = dict(kv.split("=") for kv in param_str.split(",") if kv)
+    return {k: to_number(v) for k, v in str_dict.items()}
+
+
 def get_parser() -> argparse.ArgumentParser:
     parser = argparse.ArgumentParser(description="Run evals through the API")
     parser.add_argument(
@@ -162,8 +179,7 @@ def to_number(x: str) -> Union[int, float, str]:
         eval_spec.args.update(extra_eval_params)
 
     # If the user provided an argument to --completion_args, parse it into a dict here, to be passed to the completion_fn creation **kwargs
-    completion_args = args.completion_args.split(",")
-    additional_completion_args = {k: v for k, v in (kv.split("=") for kv in completion_args if kv)}
+    additional_completion_args = parse_key_value_args(args.completion_args)
 
     completion_fns = args.completion_fn.split(",")
     completion_fn_instances = [

evals/cli/oaieval_test.py

@@ -0,0 +1,18 @@
+import os
+
+os.environ.setdefault("OPENAI_API_KEY", "dummy")
+
+from evals.cli.oaieval import parse_key_value_args
+from evals.registry import Registry
+
+
+def test_parse_key_value_args_coerces_numbers() -> None:
+    parsed = parse_key_value_args("temperature=0.5,max_tokens=10,label=test")
+
+    assert parsed == {"temperature": 0.5, "max_tokens": 10, "label": "test"}
+
+
+def test_registry_make_completion_fn_passes_completion_args_to_chat_models() -> None:
+    completion_fn = Registry().make_completion_fn("gpt-3.5-turbo", temperature=0.5)
+
+    assert completion_fn.extra_options == {"temperature": 0.5}

evals/completion_fns/openai.py

@@ -94,7 +94,10 @@ def __init__(
         self.api_base = api_base
         self.api_key = api_key
         self.n_ctx = n_ctx
-        self.extra_options = extra_options
+        self.extra_options = {
+            **(extra_options or {}),
+            **{key: value for key, value in kwargs.items() if key != "registry"},
+        }
 
     def __call__(
         self,
@@ -139,12 +142,16 @@ def __init__(
         api_key: Optional[str] = None,
         n_ctx: Optional[int] = None,
         extra_options: Optional[dict] = {},
+        **kwargs,
     ):
         self.model = model
         self.api_base = api_base
         self.api_key = api_key
         self.n_ctx = n_ctx
-        self.extra_options = extra_options
+        self.extra_options = {
+            **(extra_options or {}),
+            **{key: value for key, value in kwargs.items() if key != "registry"},
+        }
 
     def __call__(
         self,

evals/completion_fns/openai_test.py

@@ -0,0 +1,13 @@
+from evals.completion_fns.openai import OpenAIChatCompletionFn, OpenAICompletionFn
+
+
+def test_openai_chat_completion_fn_promotes_extra_kwargs_to_extra_options() -> None:
+    completion_fn = OpenAIChatCompletionFn(model="gpt-3.5-turbo", temperature=0.5, registry=object())
+
+    assert completion_fn.extra_options == {"temperature": 0.5}
+
+
+def test_openai_completion_fn_promotes_extra_kwargs_to_extra_options() -> None:
+    completion_fn = OpenAICompletionFn(model="text-davinci-003", temperature=0.5, registry=object())
+
+    assert completion_fn.extra_options == {"temperature": 0.5}