this is a branch for feature COT

examples/cotprompt/README.md

@@ -0,0 +1,47 @@
+# Chain-of-Thought Prompting Elicits Reasoning in Large Language Models
+
+## Overview
+
+This example contains code to implement the basic COT in the paper [Chain-of-Thought Prompting Elicits Reasoning in Large Language Models](https://arxiv.org/abs/2201.11903) by Wei, et al. 
+
+## Running the Example
+
+For terminal/CLI usage: Input and output are in the terminal window
+
+```bash
+cd examples/cotprompt
+python run_cli.py
+```
+
+An example input and output(by GPT-4o-mini).
+
+```
+> Input: Which is bigger, 9.9 or 9.11?
+> Reasoning output: 
+
+1. **Analyze the Question**: The question asks which of the two numbers, 9.9 or 9.11, is larger. To answer this, we must understand the place values and numerical representation of both numbers.
+
+2. **Consider the Values**: 9.9 can be broken down into 9 plus 0.9, while 9.11 can be broken down into 9 plus 0.11. This shows that both numbers start with the same whole number part (9).
+
+3. **Compare the Decimal Parts**: We need to compare the decimal portions: 0.9 and 0.11. The number 0.9 can also be expressed as 0.90 for comparison purposes, thus making it easier to analyze alongside 0.11.
+
+4. **Evaluate the Decimal Comparison**: Now we compare the decimals: 0.90 (equivalent to 0.9) and 0.11. Here, we observe that 0.90 is greater than 0.11 because in the tenths place, 0.90 has a 9 compared to 0.11’s 1, which indicates a larger value.
+
+5. **Conclude Comparison Process**: Since both numbers share the same whole number part of 9, the larger value will depend solely on the larger decimal part, which we have determined through comparison.
+
+At this point, we have thoroughly analyzed the components of each number and identified that the key to determining which is larger lies in the comparison of their decimal parts.
+
+>Final answer: 9.9 is bigger than 9.11.
+```
+
+## Citation
+```
+@article{wei2022chain,
+  title={Chain-of-thought prompting elicits reasoning in large language models},
+  author={Wei, Jason and Wang, Xuezhi and Schuurmans, Dale and Bosma, Maarten and Xia, Fei and Chi, Ed and Le, Quoc V and Zhou, Denny and others},
+  journal={Advances in neural information processing systems},
+  volume={35},
+  pages={24824--24837},
+  year={2022}
+}
+```

examples/cotprompt/agent/cot_agent/cot_conclude.py

@@ -0,0 +1,57 @@
+from omagent_core.engine.worker.base import BaseWorker
+from omagent_core.utils.registry import registry
+from omagent_core.models.llms.openai_gpt import OpenaiGPTLLM
+from omagent_core.models.llms.schemas import Message, Content
+from omagent_core.models.llms.base import BaseLLMBackend
+from omagent_core.utils.logger import logging
+from pathlib import Path
+
+CURRENT_PATH = Path(__file__).parents[0]
+
+@registry.register_worker()
+class COTConclusion(BaseLLMBackend, BaseWorker):
+    """根据问题和推理步骤生成最终答案。"""
+
+    llm: OpenaiGPTLLM
+
+    def _run(self, user_question:str, reasoning_result:str, *args, **kwargs):
+        # 从工作流变量中提取问题和推理结果
+        # user_question = kwargs['user_question']
+        # reasoning_result = kwargs['reasoning_result']
+
+        # 生成最终答案的 prompt
+        final_answer_prompt = self.generate_final_answer_prompt(user_question, reasoning_result)
+
+        # 调用 LLM 生成最终答案
+        final_answer = self.call_llm_with_prompt(final_answer_prompt)
+
+        # 将最终答案存储到工作流变量中
+        self.stm(self.workflow_instance_id)['final_answer'] = final_answer
+
+        self.callback.send_answer(self.workflow_instance_id, msg=final_answer)
+
+        return {'final_answer': final_answer}
+
+    def generate_final_answer_prompt(self, question: str, reasoning_result: str):
+        """生成最终答案的 prompt"""
+
+        return f"""
+        You have completed the reasoning process. Based on the following question and reasoning result, generate a concise and clear final answer.
+
+        Question: {question}
+        Reasoning steps: {reasoning_result}
+
+        Please provide the final answer:
+        """
+
+    def call_llm_with_prompt(self, prompt: str):
+        """调用 LLM 生成最终答案"""
+        chat_message = []
+
+        # Add text question as first message
+        chat_message.append(Message(role="user", message_type='text', content=prompt))
+
+        #print("chat_message++++++++++++++++++++++++++++",chat_message)
+
+        response = self.llm.generate(chat_message)
+        return response['choices'][0]['message']['content']

examples/cotprompt/agent/cot_agent/cot_input_interface.py

@@ -0,0 +1,21 @@
+from omagent_core.engine.worker.base import BaseWorker
+from omagent_core.utils.registry import registry
+from omagent_core.utils.logger import logging
+from pathlib import Path
+
+CURRENT_PATH = Path(__file__).parents[0]
+
+@registry.register_worker()
+class COTInputInterface(BaseWorker):
+    """处理用户输入问题并将其存储到工作流变量中。"""
+
+    def _run(self, *args, **kwargs):
+        # 读取用户输入的问题
+        user_input = self.input.read_input(workflow_instance_id=self.workflow_instance_id, input_prompt='请输入你的问题:')
+
+        question = user_input['messages'][-1]['content'][0]['data']
+
+        # 将问题存储到工作流的变量中
+        self.stm(self.workflow_instance_id)['user_question'] = question
+
+        return {'user_question': question}

examples/cotprompt/agent/cot_agent/cot_reasoning.py

@@ -0,0 +1,64 @@
+from omagent_core.models.llms.openai_gpt import OpenaiGPTLLM
+from omagent_core.engine.worker.base import BaseWorker
+from omagent_core.utils.registry import registry
+from omagent_core.models.llms.base import BaseLLMBackend
+from omagent_core.models.llms.schemas import Message, Content
+from omagent_core.utils.logger import logging
+from pathlib import Path
+
+CURRENT_PATH = Path(__file__).parents[0]
+
+@registry.register_worker()
+class COTReasoning(BaseLLMBackend, BaseWorker):
+    """根据问题生成推理过程的中间数据，不给出答案。"""
+
+    llm: OpenaiGPTLLM
+
+    def _run(self, user_question:str,*args, **kwargs):
+        # 从工作流变量中提取问题
+        #user_question = kwargs['user_question']
+
+        # 生成推理过程的 prompt
+        reasoning_prompt = self.generate_reasoning_prompt(user_question)
+        #print("reasoning_prompt++++++++++++++++++++++++++++",reasoning_prompt)
+
+        # 调用 LLM 生成推理步骤
+        reasoning_result = self.call_llm_with_prompt(reasoning_prompt)
+
+        # 将推理结果存储到工作流变量中
+        self.stm(self.workflow_instance_id)['reasoning_result'] = reasoning_result
+
+        self.callback.send_answer(self.workflow_instance_id, msg=reasoning_result)
+
+        return {'reasoning_result': reasoning_result}
+
+    def generate_reasoning_prompt(self, question: str):
+        """生成推理过程的 prompt"""
+        return f"""
+        Please reason step by step based on the following question and provide the reasoning process:
+
+        question: {question}
+
+        Reasoning Steps:
+        1. Analyze the question and consider possible solutions.
+        2. Reason through the different aspects of the question step by step.
+        3. Only provide the reasoning process, do not give the final answer.
+
+        Please begin reasoning:
+        """
+
+    def call_llm_with_prompt(self, prompt: str):
+        """调用 LLM 生成推理步骤"""
+
+
+        chat_message = []
+
+        # Add text question as first message
+        chat_message.append(Message(role="user", message_type='text', content=prompt))
+
+        #print("chat_message++++++++++++++++++++++++++++",chat_message)
+
+        response = self.llm.generate(chat_message)
+        if response is None:
+            raise ValueError("LLM inference returned None.")
+        return response['choices'][0]['message']['content']

examples/cotprompt/compile_container.py

@@ -0,0 +1,17 @@
+# Import core modules and components
+from omagent_core.utils.container import container
+
+# Import workflow related modules
+from pathlib import Path
+from omagent_core.utils.registry import registry
+
+# Set up path and import modules
+CURRENT_PATH = root_path = Path(__file__).parents[0]
+registry.import_module()
+
+# Register required components
+container.register_callback(callback='AppCallback')
+container.register_input(input='AppInput')
+container.register_stm("RedisSTM")
+# Compile container config
+container.compile_config(CURRENT_PATH)

examples/cotprompt/configs/llms/gpt4o.yml

@@ -0,0 +1,7 @@
+name: OpenaiGPTLLM
+model_id: gpt-4o-mini
+api_key: ${env| custom_openai_key, openai_api_key}
+endpoint: ${env| custom_openai_endpoint, https://api.openai.com/v1}
+temperature: 0
+vision: true
+response_format: json_object

examples/cotprompt/configs/llms/json_res.yml

@@ -0,0 +1,6 @@
+name: OpenaiGPTLLM
+model_id: gpt-4o-mini
+api_key: ${env| custom_openai_key, openai_api_key}
+endpoint: ${env| custom_openai_endpoint, https://api.openai.com/v1}
+temperature: 0
+#vision: true

examples/cotprompt/configs/llms/text_res.yml

@@ -0,0 +1,6 @@
+name: OpenaiGPTLLM
+model_id: gpt-4o-mini
+api_key: ${env| custom_openai_key, openai_api_key}
+endpoint: ${env| custom_openai_endpoint, https://api.openai.com/v1}
+temperature: 0
+response_format: text

examples/cotprompt/configs/tools/all_tools.yml

@@ -0,0 +1,10 @@
+llm: ${sub|text_res}
+tools:
+    - Calculator
+    - CodeInterpreter
+    - ReadFileContent
+    - WriteFileContent
+    - ShellTool
+    - name: WebSearch
+      bing_api_key: ${env|bing_api_key, null}
+      llm: ${sub|text_res}

examples/cotprompt/configs/workers/cot_conclude.yml

@@ -0,0 +1,5 @@
+name: COTConclusion
+llm: ${sub|json_res}
+tool_manager: ${sub|all_tools}
+output_parser:
+  name: StrParser

examples/cotprompt/configs/workers/cot_input_interface.yml

@@ -0,0 +1,5 @@
+name: COTInputInterface
+llm: ${sub|json_res}
+#tool_manager: ${sub|all_tools}
+output_parser:
+  name: StrParser

examples/cotprompt/configs/workers/cot_reasoning.yaml

@@ -0,0 +1,5 @@
+name: COTReasoning
+llm: ${sub|json_res}
+#tool_manager: ${sub|all_tools}
+output_parser:
+  name: StrParser

examples/cotprompt/container.yaml

@@ -0,0 +1,84 @@
+conductor_config:
+  name: Configuration
+  base_url:
+    value: http://localhost:8080
+    description: The Conductor Server API endpoint
+    env_var: CONDUCTOR_SERVER_URL
+  auth_key:
+    value: null
+    description: The authorization key
+    env_var: AUTH_KEY
+  auth_secret:
+    value: null
+    description: The authorization secret
+    env_var: CONDUCTOR_AUTH_SECRET
+  auth_token_ttl_min:
+    value: 45
+    description: The authorization token refresh interval in minutes.
+    env_var: AUTH_TOKEN_TTL_MIN
+  debug:
+    value: false
+    description: Debug mode
+    env_var: DEBUG
+connectors:
+  redis_stream_client:
+    name: RedisConnector
+    host:
+      value: localhost
+      env_var: HOST
+    port:
+      value: 6379
+      env_var: PORT
+    password:
+      value: null
+      env_var: PASSWORD
+    username:
+      value: null
+      env_var: USERNAME
+    db:
+      value: 0
+      env_var: DB
+  redis_stm_client:
+    name: RedisConnector
+    host:
+      value: localhost
+      env_var: HOST
+    port:
+      value: 6379
+      env_var: PORT
+    password:
+      value: null
+      env_var: PASSWORD
+    username:
+      value: null
+      env_var: USERNAME
+    db:
+      value: 1
+      env_var: DB
+components:
+  RedisSTM:
+    name: RedisSTM
+  DefaultCallback:
+    name: DefaultCallback
+    bot_id:
+      value: ''
+      env_var: BOT_ID
+    start_time:
+      value: 2024-11-10_20:51:13
+      env_var: START_TIME
+    folder_name:
+      value: ./running_logs/2024-11-10_20:51:13
+      env_var: FOLDER_NAME
+  AppInput:
+    name: AppInput
+  AppCallback:
+    name: AppCallback
+    bot_id:
+      value: ''
+      env_var: BOT_ID
+    start_time:
+      value: 2024-11-10_20:51:13
+      env_var: START_TIME
+    folder_name:
+      value: ./running_logs/2024-11-10_20:51:13
+      env_var: FOLDER_NAME