Code cleaning

Author: Runhui Wang

README.md

@@ -2,14 +2,14 @@
 
 ## Overview
 
-The `gsr-bench` repository provides a benchmarking and retrieval system with utilities for executing bash commands, interacting with models, and web searching. This repository is organized into several Python scripts that serve various functions, such as data retrieval, bash utilities, and model execution. 
+The `csr-bench` repository provides a benchmarking and retrieval system with utilities for executing bash commands, interacting with models, and web searching. This repository is organized into several Python scripts that serve various functions, such as data retrieval, bash utilities, and model execution. 
 
 ## Repository Structure
 
 - **`main.py`**: The entry point of the repository where the core functionality is implemented.
 - **`docker_setup.sh`**: A shell script to set up the environment and conduct experiments using Docker.
 - **`requirements.txt`**: A list of Python dependencies required to run the project.
-- **`gsr/`**: The main module directory containing various Python utilities:
+- **`csr/`**: The main module directory containing various Python utilities:
     - `bash_utils.py`: Functions related to executing and handling bash commands.
     - `retriever.py`: Handles data and information retrieval from Github Issues.
     - `model.py`: Manages machine learning model-related operations.
@@ -58,19 +58,19 @@ python main.py
 
 ### Bash Utilities
 
-The `gsr/bash_utils.py` module provides a set of utilities to execute and interact with bash commands programmatically. To use this, import the necessary functions into your script:
+The `csr/bash_utils.py` module provides a set of utilities to execute and interact with bash commands programmatically. To use this, import the necessary functions into your script:
 
 ```python
-from gsr.bash_utils import your_function_here
+from csr.bash_utils import your_function_here
 ```
 
 
 ### Model Operations
 
-The `gsr/model.py` provides functions for loading, evaluating, or interacting with machine learning models. Example:
+The `csr/model.py` provides functions for loading, evaluating, or interacting with machine learning models. Example:
 
 ```python
-from gsr.model import run_model
+from csr.model import run_model
 
 model_results = run_model(model_input)
 ```

gsr-agent/utils/multi-agent-sde.py csr-agent/utils/multi-agent-sde.pygsr-agent/utils/multi-agent-sde.py renamed to csr-agent/utils/multi-agent-sde.py

@@ -18,14 +18,11 @@ def call_claude3(inputs, start, workload_size, model_id='anthropic.claude-3-haik
                        region_name='us-west-2', config=config)
     results = []
     for i in range(start, min(start+workload_size, len(inputs))):
-        # input_prompt = inputs[i]
         input_prompt = inputs[i]['input']
         if input_prompt == 'missing':
             results.append((i, 'score: 0\n'))
             continue
 
-        # print(input_prompt)
-        
         user_message = {"role": "user", "content": f"{input_prompt}"}
         messages = [user_message]
         max_tokens = 4096
@@ -39,7 +36,6 @@ def call_claude3(inputs, start, workload_size, model_id='anthropic.claude-3-haik
             }
         )
 
-        #response = brt.invoke_model(body=body, modelId=model_id)
         while True:
             try:
                 response = brt.invoke_model(body=body, modelId=model_id)
@@ -50,7 +46,6 @@ def call_claude3(inputs, start, workload_size, model_id='anthropic.claude-3-haik
                 continue
 
         response_body = json.loads(response.get('body').read())
-        # results.append((i, response_body['content'][0]['text']))
         inputs[i]['command'] = response_body['content'][0]['text']
         inputs[i]['idx'] = i
         results.append(
@@ -67,13 +62,8 @@ def inference(input):
 class LLMAgent:
     def __init__(self, name='Agent', model_name='lmsys/vicuna-7b-v1.5'):
         self.name = name
-        # self.tokenizer = AutoTokenizer.from_pretrained(model_name)
-        # self.model = AutoModelForCausalLM.from_pretrained(model_name)
 
     def query(self, prompt):
-        # inputs = self.tokenizer(prompt, return_tensors='pt')
-        # outputs = self.model.generate(**inputs, max_length=150)
-        # response = self.tokenizer.decode(outputs[0], skip_special_tokens=True)
         response = inference(prompt)
         return response
 

gsr-agent/utils/tool_usage.py csr-agent/utils/tool_usage.pygsr-agent/utils/tool_usage.py renamed to csr-agent/utils/tool_usage.py

@@ -4,7 +4,7 @@
 from botocore.exceptions import ClientError
 from botocore.config import Config
 
-from gsr.perplexity import pplx
+from csr.perplexity import pplx
 
 
 class ToolAgent:
@@ -22,12 +22,6 @@ def __init__(self, model_id, tool_config, region_name='us-west-2', max_attempts=
         # self.tools = dict()
         logging.basicConfig(level=logging.INFO)
 
-    # def get_top_song(self, call_sign):
-    #     """Returns the most popular song for the requested station."""
-    #     if call_sign == 'WZPZ':
-    #         return "Elemental Hotel", "8 Storey Hike"
-    #     else:
-    #         raise StationNotFoundError(f"Station {call_sign} not found.")
 
     def pplx(self, message):
         return pplx(message)
@@ -45,18 +39,14 @@ def generate_text(self, input_text):
         output_message = response['output']['message']
         messages.append(output_message)
         stop_reason = response['stopReason']
-        # print(response)
 
         if stop_reason == 'tool_use':
             tool_requests = output_message['content']
-            # print('## Tool Requests ##', tool_requests)
             for tool_request in tool_requests:
                 if 'toolUse' in tool_request:
                     tool = tool_request['toolUse']
                     self.logger.info("Requesting tool %s. Request: %s", tool['name'], tool['toolUseId'])
                     try:
-                        # song, artist = self.get_top_song(tool['input']['sign'])
-                        # tool_result = {"toolUseId": tool['toolUseId'], "content": [{"json": {"song": song, "artist": artist}}]}
                         analysis = self.pplx(tool['input']['log_msg'])
                         tool_result = {"toolUseId": tool['toolUseId'], "content": [{"text": analysis}]}
                     except Exception as err:
@@ -71,24 +61,9 @@ def generate_text(self, input_text):
                     )
                     output_message = response['output']['message']
 
-        # for content in output_message['content']:
-        #     print(json.dumps(content, indent=4))
         print(output_message['content'][0]['text'])
 
-#     def main(self):
-#         input_text = """Traceback (most recent call last):
-#   File "<stdin>", line 1, in <module>
-# ModuleNotFoundError: No module named 'esm'"""
-
-#         print(f"Question: {input_text}")
-#         try:
-#             self.generate_text(input_text)
-#         except ClientError as err:
-#             message = err.response['Error']['Message']
-#             self.logger.error("A client error occurred: %s", message)
-#             print(f"A client error occurred: {message}")
-#         else:
-#             print(f"Finished generating text with model {self.model_id}")
+
     def query(self, msg):
         try:
             self.generate_text(msg)

gsr-agent/utils/vLLM.py csr-agent/utils/vLLM.pygsr-agent/utils/vLLM.py renamed to csr-agent/utils/vLLM.py

@@ -1,6 +1,6 @@
 import pexpect
 import re
-from gsr.model import CoreAgent
+from csr.model import CoreAgent
 
 
 sys_prompt = """Given the execution output, decide if the command is successful or not.
@@ -176,21 +176,21 @@ def close(self):
     executor = CommandExecutor(timeout=30)
 
     # Assuming you need to activate an environment
-    execution_ret = executor.execute_cmd('python -m venv venv', directory='/home/yijia/git-bench/data/gsr_data/Repo/')
+    execution_ret = executor.execute_cmd('python -m venv venv', directory='./data/csr_data/Repo/')
     # print(execution_ret)
 
     # Execute commands in the activated environment
-    execution_ret = executor.execute_cmd('source venv/bin/activate', directory='/home/yijia/git-bench/data/gsr_data/Repo/')
+    execution_ret = executor.execute_cmd('source venv/bin/activate', directory='./data/csr_data/Repo/')
     # print(execution_ret)
 
     # Optionally execute a command in a specific directory
-    execution_ret = executor.execute_cmd('pip install -r requirements.txt', directory='/home/yijia/git-bench/data/gsr_data/Repo/')
+    execution_ret = executor.execute_cmd('pip install -r requirements.txt', directory='./data/csr_data/Repo/')
     # print(execution_ret)
 
-    execution_ret = executor.execute_cmd('which pip', directory='/home/yijia/git-bench/data/gsr_data/Repo/')
+    execution_ret = executor.execute_cmd('which pip', directory='./data/csr_data/Repo/')
     # print(execution_ret)
 
-    execution_ret = executor.execute_cmd('ABCBACBAC', directory='/home/yijia/git-bench/data/gsr_data/Repo/')
+    execution_ret = executor.execute_cmd('ABCBACBAC', directory='./data/csr_data/Repo/')
     # print(execution_ret)
 
 

gsr/__init__.py csr/__init__.pygsr/__init__.py renamed to csr/__init__.py

@@ -50,13 +50,13 @@
 
     "Llama_3.1_8b_Instruct": "meta.llama3-1-8b-instruct-v1:0",
     "Llama_3.1_70b_Instruct": "meta.llama3-1-70b-instruct-v1:0",
-    # "Llama_3.1_405B_Instruct": "meta.llama3-1-405b-instruct-v1:0",
+    "Llama_3.1_405B_Instruct": "meta.llama3-1-405b-instruct-v1:0",
 
-    # "Mistral_7B_Instruct": "mistral.mistral-7b-instruct-v0:2",
-    # "Mixtral_8X7B_Instruct": "mistral.mixtral-8x7b-instruct-v0:1",
+    "Mistral_7B_Instruct": "mistral.mistral-7b-instruct-v0:2",
+    "Mixtral_8X7B_Instruct": "mistral.mixtral-8x7b-instruct-v0:1",
     "Mistral_Large": "mistral.mistral-large-2402-v1:0",
     "Mistral_Large_2": "mistral.mistral-large-2407-v1:0",
-    # "Mistral_Small": "mistral.mistral-small-2402-v1:0",
+    "Mistral_Small": "mistral.mistral-small-2402-v1:0",
     # gpt-4o gpt-4o-mini gpt-4-turbo
     "GPT_4o": "gpt-4o",
     "GPT_4o_Mini": "gpt-4o-mini",