Skip to content

Commit

Permalink
add double check logic before execution
Browse files Browse the repository at this point in the history
  • Loading branch information
@DoraDong-2023
DoraDong-2023 committed Dec 26, 2023
1 parent 67b319e commit a12be41
Show file tree
Hide file tree
Showing 2 changed files with 101 additions and 57 deletions.
150 changes: 93 additions & 57 deletions src/deploy/inference_dialog_server.py
View file
Original file line number Diff line number Diff line change
Expand Up @@ -53,7 +53,7 @@
from inference.retriever_finetune_inference import ToolRetriever
from deploy.utils import dataframe_to_markdown, convert_image_to_base64
from prompt.parameters import prepare_parameters_prompt
from prompt.summary import prepare_summary_prompt
from prompt.summary import prepare_summary_prompt, prepare_summary_prompt_full

basic_types = ['str', 'int', 'float', 'bool', 'list', 'dict', 'tuple', 'set', 'List', 'Dict']
basic_types.extend(['_AvailShapes']) # extend for squidpy `shape` type
Expand Down Expand Up @@ -200,6 +200,35 @@ def save_decoded_file(raw_file):
pass
return filename

def correct_bool_values(optional_param):
"""
Convert boolean values from lowercase (true, false) to uppercase (True, False).
:param optional_param: The dictionary containing the optional parameters.
:return: The modified dictionary with corrected boolean values.
"""
for key, value in optional_param.items():
if 'optional' in value and isinstance(value['optional'], bool):
value['optional'] = str(value['optional'])
if 'optional_value' in value and isinstance(value['optional_value'], bool):
value['optional_value'] = str(value['optional_value'])
return optional_param

def convert_bool_values(optional_param):
"""
Convert 'true' and 'false' in the 'optional' and 'optional_value' fields
to 'True' and 'False' respectively.
:param optional_param: The dictionary containing the optional parameters.
:return: The modified dictionary with converted boolean values.
"""
for key, value in optional_param.items():
if 'optional' in value and isinstance(value['optional'], str):
value['optional'] = value['optional'].capitalize()
if 'optional_value' in value and isinstance(value['optional_value'], str):
value['optional_value'] = value['optional_value'].capitalize()
return optional_param

class Model:
def __init__(self):
logging.info("Initializing...")
Expand Down Expand Up @@ -712,6 +741,8 @@ def run_pipeline(self, user_input, lib, top_k=3, files=[],conversation_started=T
self.run_select_basic_params(user_input)
elif self.user_states == "run_pipeline_after_entering_params":
self.run_pipeline_after_entering_params(user_input)
elif self.user_states == "run_pipeline_after_doublechecking_execution_code":
self.run_pipeline_after_doublechecking_execution_code(user_input)
def run_pipeline_after_ambiguous(self,user_input):
logging.info('==>run_pipeline_after_ambiguous')
user_input = user_input.strip()
Expand Down Expand Up @@ -1136,68 +1167,68 @@ def run_pipeline_after_entering_params(self, user_input):
"class_selected_params":class_selected_params})
logging.info('==>api_params_list: %s', json.dumps(api_params_list))
# add optional cards
optional_param = {key: value for key, value in self.API_composite[api_name]['Parameters'].items() if not value.get('optional', False)}
optional_param = {key: value for key, value in self.API_composite[api_name]['Parameters'].items() if value['optional']}
logging.info('==>optional_param: %s', json.dumps(optional_param))
if optional_param:
[callback.on_agent_action(block_id="optional-"+str(self.indexxxx),task=optional_param,task_title="Executed code",) for callback in self.callbacks]
self.indexxxx+=1
"""if True:
[callback.on_agent_action(block_id="optional-"+str(self.indexxxx),task={
"obs": {
"type": "Optional[Union[str, Iterable[str]]]",
"default": "None",
"optional": True,
"description": "Labels' keys in `adata_ref.obs` which need to be mapped to `adata.obs`\n(inferred for observation of `adata`).",
"optional_value": False
},
"embedding_method": {
"type": "Union[str, Iterable[str]]",
"default": "('umap', 'pca')",
"optional": True,
"description": "Embeddings in `adata_ref` which need to be mapped to `adata`.\nThe only supported values are 'umap' and 'pca'.",
"optional_value": False
},
"labeling_method": {
"type": "str",
"default": "knn",
"optional": True,
"description": "The method to map labels in `adata_ref.obs` to `adata.obs`.\nThe only supported value is 'knn'.",
"optional_value": False
},
"neighbors_key": {
"type": "Optional[str]",
"default": "None",
"optional": True,
"description": "If not specified, ingest looks adata_ref.uns['neighbors']\nfor neighbors settings and adata_ref.obsp['distances'] for\ndistances (default storage places for pp.neighbors).\nIf specified, ingest looks adata_ref.uns[neighbors_key] for\nneighbors settings and\nadata_ref.obsp[adata_ref.uns[neighbors_key]['distances_key']] for distances.",
"optional_value": False
},
"inplace": {
"type": "bool",
"default": "True",
"optional": True,
"description": "Only works if `return_joint=False`.\nAdd labels and embeddings to the passed `adata` (if `True`)\nor return a copy of `adata` with mapped embeddings and labels.",
"optional_value": False
}
},task_title="Executed code",) for callback in self.callbacks]
self.indexxxx+=1"""
execution_code = self.executor.generate_execution_code(api_params_list)
logging.info('==>execution_code: %s', execution_code)
print('len(optional_param)', len(optional_param))
[callback.on_tool_start() for callback in self.callbacks]
[callback.on_tool_end() for callback in self.callbacks]
if False:
if len(optional_param)>0:
print('producing optional param card')
[callback.on_agent_action(block_id="optional-"+str(self.indexxxx),task=convert_bool_values(correct_bool_values(optional_param)),task_title="Executed code",) for callback in self.callbacks]
self.indexxxx+=1
else:
pass
# TODO: real time adjusting execution_code according to optionalcard
self.execution_code = self.executor.generate_execution_code(api_params_list)
logging.info('==>execution_code: %s', self.execution_code)
[callback.on_tool_start() for callback in self.callbacks]
[callback.on_tool_end() for callback in self.callbacks]
[callback.on_agent_action(block_id="code-"+str(self.indexxxx),task=execution_code,task_title="Executed code",) for callback in self.callbacks]
[callback.on_agent_action(block_id="code-"+str(self.indexxxx),task=self.execution_code,task_title="Executed code",) for callback in self.callbacks]
self.indexxxx+=1
# LLM response
summary_prompt = prepare_summary_prompt_full(user_input, self.predicted_api_name, self.API_composite[self.predicted_api_name]['description'], self.API_composite[self.predicted_api_name]['Parameters'],self.API_composite[self.predicted_api_name]['Returns'], self.execution_code)
response, _ = LLM_response(self.llm, self.tokenizer, summary_prompt, history=[], kwargs={})
[callback.on_agent_action(block_id="log-"+str(self.indexxxx),task=response,task_title=f"Task summary before execution",) for callback in self.callbacks]
self.indexxxx+=1
[callback.on_agent_action(block_id="log-"+str(self.indexxxx),task="Could you confirm whether this task is what you aimed for, and the code should be executed? Please enter y/n.\nIf you press n, then we will re-direct to the parameter input step",task_title=f"Double Check ",) for callback in self.callbacks]
self.indexxxx+=1
execution_code_list = execution_code.split('\n')
self.last_user_states = self.user_states
self.user_states = "run_pipeline_after_doublechecking_execution_code"
self.save_state()

def run_pipeline_after_doublechecking_execution_code(self, user_input):
# if check, back to the last iteration and status
if user_input in ['y', 'n']:
if user_input == 'n':
print('input n')
self.last_user_states = self.user_states
#self.user_states = "initial"
self.user_states = "run_pipeline_after_doublechecking_API_selection" #TODO: check if exist issue
[callback.on_tool_start() for callback in self.callbacks]
[callback.on_tool_end() for callback in self.callbacks]
[callback.on_agent_action(block_id="log-"+str(self.indexxxx),task="We will redirect to the parameters input",task_title=f"Re-enter the parameters",) for callback in self.callbacks]
self.indexxxx+=1
self.save_state()
self.run_pipeline_after_doublechecking_API_selection('y')
return
else:
print('input y')
pass
else:
print('input not y or n')
[callback.on_tool_start() for callback in self.callbacks]
[callback.on_tool_end() for callback in self.callbacks]
[callback.on_agent_action(block_id="log-"+str(self.indexxxx),task="The input was not y or n, please enter the correct value.",task_title=f"Index Error",) for callback in self.callbacks]
self.indexxxx+=1
self.save_state()
# user_states didn't change
return
# else, continue
execution_code_list = self.execution_code.split('\n')
logging.info('execute and obtain figures')
self.plt_status = plt.get_fignums()
#self.hide_streams()
# execute and obtain figures
#self.restore_streams()
#content = self.buf1.getvalue()
#logging.info('content, %s', content)
#content = self.buf2.getvalue()
#logging.info('content, %s', content)
temp_output_file = "./sub_process_execution.txt"
#open(temp_output_file, 'w').close()
process = multiprocessing.Process(target=self.run_pipeline_execution_code_list, args=(execution_code_list, temp_output_file))
process.start()
while process.is_alive():
Expand All @@ -1208,7 +1239,6 @@ def run_pipeline_after_entering_params(self, user_input):
print('accumulated_output', accumulated_output)
[callback.on_agent_action(block_id="log-"+str(self.indexxxx),task=accumulated_output,task_title="Executing results",) for callback in self.callbacks]
self.indexxxx+=1
#os.remove(temp_output_file)

with open("./tmp/tmp_output_run_pipeline_execution_code_list.txt", 'r') as file:
output_str = file.read()
Expand Down Expand Up @@ -1314,13 +1344,19 @@ def run_pipeline_after_entering_params(self, user_input):
self.indexxxx+=1
[callback.on_agent_action(block_id="log-"+str(self.indexxxx),task="Splitting the returned tuple variable into individual variables",task_title="Executed results [Success]",) for callback in self.callbacks]
self.indexxxx+=1
else:
pass
else:
pass
logging.info("Show current variables in namespace:")
logging.info(json.dumps(list(self.executor.variables.keys())))
new_str = []
for i in self.executor.execute_code:
new_str.append({"code":i['code'],"execution_results":i['success']})
logging.info("Currently all executed code: %s", json.dumps(new_str))
filename = f"./tmp/sessions/{str(self.session_id)}_environment.pkl"
self.last_user_states = self.user_states
self.user_states = "initial"
self.executor.save_environment(filename)
self.save_state()
def run_pipeline_execution_code_list(self, execution_code_list, output_file):
Expand Down
8 changes: 8 additions & 0 deletions src/prompt/summary.py
View file
Original file line number Diff line number Diff line change
Expand Up @@ -5,3 +5,11 @@ def prepare_summary_prompt(user_query, api_function, api_description, parameters
and includes the selected API function `{api_function}` which functionality is `{api_description}` with its parameters `{parameters_description}`. The returned variable is `{return_information}`
Please use the template as `The task is ..., we solved it by ...`. The response should be in three sentences
"""

def prepare_summary_prompt_full(user_query, api_function, api_description, parameters_description,return_information,execution_code):
return f"""
Help to summarize the task with its solution in layman tone, it should be understandable by non professional programmers. Starts from description of the user query `{user_query}`
and includes the selected API function `{api_function}` which functionality is `{api_description}` with its parameters `{parameters_description}`. The returned variable is `{return_information}`. The generated code is `{execution_code}`.
Please use the template as `The task is ..., we solved it by ...`. The response should be in four sentences. Additionally, the interpretation encompasses explanations for the parameters utilized in the generated code.
"""

0 comments on commit a12be41

Please sign in to comment.