Fix default_sampling_params and simple_workflow (#373)

Author: chenyushuo

trinity/common/models/vllm_model.py

@@ -62,14 +62,14 @@ def __init__(
             )
         self.default_sampling_params = vllm.SamplingParams(
             n=1,
-            temperature=0.0,
+            temperature=config.temperature,
             max_tokens=config.max_response_tokens,
             min_tokens=config.min_response_tokens,
             truncate_prompt_tokens=config.max_prompt_tokens,
             skip_special_tokens=True,
             include_stop_str_in_output=False,
             output_kind=RequestOutputKind.FINAL_ONLY,
-            logprobs=0,
+            logprobs=config.logprobs,
             ignore_eos=config.ignore_eos,
         )
         self.enable_thinking = config.enable_thinking

trinity/common/workflows/workflow.py

@@ -190,13 +190,7 @@ def process_messages_to_experience(self, messages, reward, info={}) -> Experienc
         return experience
 
 
-@WORKFLOWS.register_module("simple_workflow")
-class SimpleWorkflow(Workflow):
-    """A workflow for simple single-round task."""
-
-    can_reset: bool = True
-    can_repeat: bool = True
-
+class BaseSimpleWorkflow(Workflow):
     def __init__(
         self,
         *,
@@ -246,6 +240,14 @@ def format_messages(self):
             messages.append({"role": "assistant", "content": self.reply_prefix})
         return messages
 
+
+@WORKFLOWS.register_module("simple_workflow")
+class SimpleWorkflow(BaseSimpleWorkflow):
+    """A workflow for simple single-round task."""
+
+    can_reset: bool = True
+    can_repeat: bool = True
+
     def run(self) -> List[Experience]:
         # TODO: Optimize the generate function
         messages = self.format_messages()
@@ -272,7 +274,7 @@ def run(self) -> List[Experience]:
 
 
 @WORKFLOWS.register_module("async_simple_workflow")
-class AsyncSimpleWorkflow(Workflow):
+class AsyncSimpleWorkflow(BaseSimpleWorkflow):
     is_async: bool = True
 
     async def run_async(self) -> List[Experience]: