Bump default gpt-3.5-turbo-instruct max tokens to 256, refs #284

llm/default_plugins/openai_models.py

@@ -23,7 +23,7 @@ def register_models(register):
     register(Chat("gpt-4"), aliases=("4", "gpt4"))
     register(Chat("gpt-4-32k"), aliases=("4-32k",))
     register(
-        Completion("gpt-3.5-turbo-instruct"),
+        Completion("gpt-3.5-turbo-instruct", default_max_tokens=256),
         aliases=("3.5-instruct", "chatgpt-instruct"),
     )
     # Load extra models
@@ -126,6 +126,8 @@ class Chat(Model):
     key_env_var = "OPENAI_API_KEY"
     can_stream: bool = True
 
+    default_max_tokens = None
+
     class Options(llm.Options):
         temperature: Optional[float] = Field(
             description=(
@@ -280,6 +282,8 @@ class Chat(Model):
 
     def build_kwargs(self, prompt):
         kwargs = dict(not_nulls(prompt.options))
+        if "max_tokens" not in kwargs and self.default_max_tokens is not None:
+            kwargs["max_tokens"] = self.default_max_tokens
         if self.api_base:
             kwargs["api_base"] = self.api_base
         if self.api_type:
@@ -301,6 +305,10 @@ class Chat(Model):
 
 
 class Completion(Chat):
+    def __init__(self, *args, default_max_tokens=None, **kwargs):
+        super().__init__(*args, **kwargs)
+        self.default_max_tokens = default_max_tokens
+
     def __str__(self):
         return "OpenAI Completion: {}".format(self.model_id)
 

tests/test_llm.py

@@ -313,6 +313,15 @@ def test_openai_completion(mocked_openai_completion, user_path):
     )
     assert result.exit_code == 0
     assert result.output == "\n\nThis is indeed a test\n"
+
+    # Should have requested 256 tokens
+    assert json.loads(mocked_openai_completion.last_request.text) == {
+        "model": "gpt-3.5-turbo-instruct",
+        "prompt": "Say this is a test",
+        "stream": False,
+        "max_tokens": 256,
+    }
+
     # Check it was logged
     rows = list(log_db["responses"].rows)
     assert len(rows) == 1