github/ColossalAI
1
0
Fork 0
mirror of https://github.com/hpcaitech/ColossalAI.git synced 2025-09-01 01:06:00 +00:00
Code Issues Packages Projects Releases Wiki Activity

[hotfix] fix typo s/keywrods/keywords etc. (#5429)

Browse Source
This commit is contained in:
digger yu
2024-03-12 11:25:16 +08:00
committed by GitHub
parent da885ed540
commit 385e85afd4
17 changed files with 25 additions and 25 deletions
Download Patch File Download Diff File Expand all files Collapse all files

8
applications/ColossalQA/colossalqa/chain/retrieval_qa/base.py
View File

@@ -117,8 +117,8 @@ class CustomBaseRetrievalQA(BaseRetrievalQA):
) = copy.deepcopy(buffered_history_backup), copy.deepcopy(summarized_history_temp_backup)
# if rejection_trigger_keywords is not given, return the response from LLM directly
rejection_trigger_keywrods = inputs.get('rejection_trigger_keywrods', [])
answer = answer if all([rej not in answer for rej in rejection_trigger_keywrods]) else None
rejection_trigger_keywords = inputs.get('rejection_trigger_keywords', [])
answer = answer if all([rej not in answer for rej in rejection_trigger_keywords]) else None
if answer is None:
answer = inputs.get('rejection_answer', "抱歉，根据提供的信息无法回答该问题。")
if self.combine_documents_chain.memory is not None:
@@ -161,8 +161,8 @@ class CustomBaseRetrievalQA(BaseRetrievalQA):
input_documents=docs, question=question, callbacks=_run_manager.get_child(), **kwargs
)
# if rejection_trigger_keywords is not given, return the response from LLM directly
rejection_trigger_keywrods = inputs.get('rejection_trigger_keywrods', [])
answer = answer if all([rej not in answer for rej in rejection_trigger_keywrods]) or len(rejection_trigger_keywrods)==0 else None
rejection_trigger_keywords = inputs.get('rejection_trigger_keywords', [])
answer = answer if all([rej not in answer for rej in rejection_trigger_keywords]) or len(rejection_trigger_keywords)==0 else None
if answer is None:
answer = inputs.get('rejection_answer', "抱歉，根据提供的信息无法回答该问题。")
self.combine_documents_chain.memory.save_context({"question": question}, {"output": answer})

6
applications/ColossalQA/colossalqa/prompt/prompt.py
View File

@@ -75,7 +75,7 @@ Assistant: 我认识一个叫张三的人
# Below are English retrieval qa prompts
_EN_RETRIEVAL_QA_PROMPT = """[INST] <<SYS>>Always answer as helpfully as possible, while being safe. Your answers should not include any harmful, unethical, racist, sexist content.
If the answer cannot be infered based on the given context, please say "I cannot answer the question based on the information given.".<</SYS>>
If the answer cannot be inferred based on the given context, please say "I cannot answer the question based on the information given.".<</SYS>>
Use the context and chat history to answer the question.
context:
@@ -97,8 +97,8 @@ Chat history:
Human: I have a friend, Mike. Do you know him?
Assistant: Yes, I know a person named Mike
sentence: What's his favorate food?
disambiguated sentence: What's Mike's favorate food?
sentence: What's his favorite food?
disambiguated sentence: What's Mike's favorite food?
[/INST]
Chat history:
{chat_history}

2
applications/ColossalQA/colossalqa/retrieval_conversation_en.py
View File

@@ -80,7 +80,7 @@ class EnglishRetrievalConversation:
self.retrieval_chain.run(
query=user_input,
stop=[self.memory.human_prefix + ": "],
rejection_trigger_keywrods=["cannot answer the question"],
rejection_trigger_keywords=["cannot answer the question"],
rejection_answer="Sorry, this question cannot be answered based on the information provided.",
).split("\n")[0],
self.memory,

2
applications/ColossalQA/colossalqa/retrieval_conversation_universal.py
View File

@@ -103,7 +103,7 @@ class UniversalRetrievalConversation:
break
data_name = input("Enter a short description of the data:")
separator = input(
"Enter a separator to force separating text into chunks, if no separator is given, the defaut separator is '\\n\\n', press ENTER directly to skip:"
"Enter a separator to force separating text into chunks, if no separator is given, the default separator is '\\n\\n', press ENTER directly to skip:"
)
separator = separator if separator != "" else "\n\n"
retriever_data = DocumentLoader([[file, data_name.replace(" ", "_")]]).all_data

2
applications/ColossalQA/colossalqa/retrieval_conversation_zh.py
View File

@@ -87,7 +87,7 @@ class ChineseRetrievalConversation:
query=user_input,
stop=["</答案>"],
doc_prefix="支持文档",
rejection_trigger_keywrods=["无法回答该问题"],
rejection_trigger_keywords=["无法回答该问题"],
rejection_answer="抱歉，根据提供的信息无法回答该问题。",
).split("\n")[0],
self.memory,

2
applications/ColossalQA/examples/retrieval_conversation_chatgpt.py
View File

@@ -61,7 +61,7 @@ if __name__ == "__main__":
information_retriever.add_documents(docs=documents, cleanup="incremental", mode="by_source", embedding=embedding)
prompt_template = """Always answer as helpfully as possible, while being safe. Your answers should not include any harmful, unethical, racist, sexist, toxic, dangerous, or illegal content. Please ensure that your responses are socially unbiased and positive in nature.
If the answer cannot be infered based on the given context, please don't share false information.
If the answer cannot be inferred based on the given context, please don't share false information.
Use the context and chat history to respond to the human's input at the end or carry on the conversation. You should generate one response only. No following up is needed.
context:

4
applications/ColossalQA/examples/retrieval_conversation_en.py
View File

@@ -67,7 +67,7 @@ if __name__ == "__main__":
break
data_name = input("Enter a short description of the data:")
separator = input(
"Enter a separator to force separating text into chunks, if no separator is given, the defaut separator is '\\n\\n'. Note that"
"Enter a separator to force separating text into chunks, if no separator is given, the default separator is '\\n\\n'. Note that"
+ "we use neural text spliter to split texts into chunks, the seperator only serves as a delimiter to force split long passage into"
+ " chunks before passing to the neural network. Press ENTER directly to skip:"
)
@@ -112,7 +112,7 @@ if __name__ == "__main__":
agent_response = retrieval_chain.run(
query=user_input,
stop=["Human: "],
rejection_trigger_keywrods=EN_RETRIEVAL_QA_TRIGGER_KEYWORDS,
rejection_trigger_keywords=EN_RETRIEVAL_QA_TRIGGER_KEYWORDS,
rejection_answer=EN_RETRIEVAL_QA_REJECTION_ANSWER,
)
agent_response = agent_response.split("\n")[0]

2
applications/ColossalQA/examples/retrieval_conversation_en_customer_service.py
View File

@@ -142,7 +142,7 @@ if __name__ == "__main__":
agent_response = retrieval_chain.run(
query=user_input,
stop=["Human: "],
rejection_trigger_keywrods=EN_RETRIEVAL_QA_TRIGGER_KEYWORDS,
rejection_trigger_keywords=EN_RETRIEVAL_QA_TRIGGER_KEYWORDS,
rejection_answer=EN_RETRIEVAL_QA_REJECTION_ANSWER,
)
agent_response = agent_response.split("\n")[0]

2
applications/ColossalQA/examples/retrieval_conversation_universal.py
View File

@@ -11,7 +11,7 @@ if __name__ == '__main__':
parser.add_argument('--sql_file_path', type=str, default=None, help='path to the a empty folder for storing sql files for indexing')
args = parser.parse_args()
# Will ask for documents path in runnning time
# Will ask for documents path in running time
session = UniversalRetrievalConversation(files_en=None,
files_zh=None,
zh_model_path=args.zh_model_path, en_model_path=args.en_model_path,

2
applications/ColossalQA/examples/retrieval_conversation_zh.py
View File

@@ -107,7 +107,7 @@ if __name__ == "__main__":
query=user_input,
stop=["</答案>"],
doc_prefix="支持文档",
rejection_trigger_keywrods=ZH_RETRIEVAL_QA_TRIGGER_KEYWORDS,
rejection_trigger_keywords=ZH_RETRIEVAL_QA_TRIGGER_KEYWORDS,
rejection_answer=ZH_RETRIEVAL_QA_REJECTION_ANSWER,
)
print(f"Agent: {agent_response}")

2
applications/ColossalQA/examples/webui_demo/RAG_ChatBot.py
View File

@@ -140,7 +140,7 @@ class RAG_ChatBot:
result = self.rag_chain.run(
query=user_input,
stop=[memory.human_prefix + ": "],
rejection_trigger_keywrods=ZH_RETRIEVAL_QA_TRIGGER_KEYWORDS,
rejection_trigger_keywords=ZH_RETRIEVAL_QA_TRIGGER_KEYWORDS,
rejection_answer=ZH_RETRIEVAL_QA_REJECTION_ANSWER,
)
return result, memory