Remove useless configs (#113)

Author: HYLcool

docs/sphinx_doc/source/tutorial/example_data_functionalities.md

@@ -88,14 +88,10 @@ data_processor:
     #### new part about data active iterator
     dj_process_desc: 'Please compute difficulty scores for these math questions.'
     agent_model_name: 'qwen-max'
-    agent_model_config:
-      config_name: 'my-qwen-instruction'
-      model_type: 'dashscope_chat'
-      model_name: 'qwen2.5-72b-instruct'
     clean_strategy: 'iterative'
 ```
 
-You can write your demand description in config item `dj_process_desc`, and set the model name and configs used for the agent in config items `agent_model_name` and `agent_model_config`. Here we use Qwen2.5-72b-Instruct as our recipe managing agent. And you can set the `clean_strategy` to 'iterative' to get a better dataset.
+You can write your demand description in config item `dj_process_desc`, and set the model name used for the agent in config items `agent_model_name`. Here we use Qwen-Max as our recipe managing agent. And you can set the `clean_strategy` to 'iterative' to get a better dataset.
 
 #### Familiar with Data-Juicer
 If you are familiar with Data-Juicer, you will realize that Data-Juicer provides an operator that can calculate difficulty scores using existing LLMs. In this case, you can prepare your own Data-Juicer data processing recipe and equip this operator. For example:

examples/grpo_gsm8k_task_pipeline/gsm8k.yaml

@@ -25,10 +25,6 @@ data_processor:
     # data active iterator related
     dj_process_desc: 'Please compute difficulty scores for these math questions.'
     agent_model_name: 'qwen-max'
-    agent_model_config:
-      config_name: 'my-qwen-instruction'
-      model_type: 'dashscope_chat'
-      model_name: 'qwen2.5-72b-instruct'
     clean_strategy: 'iterative'
 
 model:

tests/data/controllers/task_parser_test.py

@@ -3,7 +3,6 @@
 import os
 import unittest
 
-import agentscope
 from agentscope.models import DashScopeChatWrapper
 from loguru import logger
 
@@ -19,15 +18,6 @@ def setUp(self) -> None:
 
         api_key = os.environ.get("OPENAI_API_KEY", None)
 
-        agentscope.init(
-            model_configs=[
-                {
-                    "config_name": "my-qwen-instruction",
-                    "model_type": "dashscope_chat",
-                    "model_name": "qwen2.5-72b-instruct",
-                },
-            ],
-        )
         self.agent = DashScopeChatWrapper(
             config_name="_",
             model_name="qwen-max",

trinity/common/config.py

@@ -137,7 +137,6 @@ class DataPipelineConfig:
         str
     ] = None  # Describe the data processing procedure without requiring users to be aware of the specific DJ parameters
     agent_model_name: Optional[str] = None
-    agent_model_config: Optional[Dict[str, Any]] = None
     clean_strategy: str = "iterative"
     min_size_ratio: Optional[float] = None
     min_priority_score: Optional[float] = 0.0

trinity/data/controllers/active_iterator.py

@@ -46,14 +46,12 @@ def __init__(
             self.pipeline_type = DataProcessorPipelineType(pipeline_type)
 
         # check if the llm agent is required
-        if self.config.agent_model_name is not None and self.config.agent_model_config is not None:
+        if self.config.agent_model_name is not None:
             # get the api key
             api_key = os.environ.get("OPENAI_API_KEY")
             # initialize the agent
-            import agentscope
             from agentscope.models import DashScopeChatWrapper
 
-            agentscope.init(model_configs=[self.config.agent_model_config])
             self.llm_agent = DashScopeChatWrapper(
                 config_name="_",
                 model_name=self.config.agent_model_name,