File size: 9,383 Bytes
fc2cb23 1e2550f e029e22 e165ea5 1e2550f e165ea5 fc2cb23 e165ea5 b409192 e165ea5 b409192 2719f21 b409192 e165ea5 b409192 e165ea5 fc2cb23 e029e22 fc2cb23 e029e22 fc2cb23 e029e22 fc2cb23 e029e22 fc2cb23 2719f21 fc2cb23 b409192 fc2cb23 e029e22 fc2cb23 b409192 fc2cb23 b409192 e029e22 4de6b1a c658776 e029e22 c658776 e029e22 |
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 266 267 268 269 270 271 272 273 274 275 276 277 278 279 |
from langchain_core.prompts import ChatPromptTemplate
# from modules.chat.langchain.utils import
from langchain_community.chat_message_histories import ChatMessageHistory
from modules.chat.base import BaseRAG
from langchain_core.prompts import PromptTemplate
from langchain.memory import ConversationBufferWindowMemory
from langchain_core.runnables.utils import ConfigurableFieldSpec
from .utils import (
CustomConversationalRetrievalChain,
create_history_aware_retriever,
create_stuff_documents_chain,
create_retrieval_chain,
return_questions,
CustomRunnableWithHistory,
BaseChatMessageHistory,
InMemoryHistory,
)
class Langchain_RAG_V1(BaseRAG):
def __init__(
self,
llm,
memory,
retriever,
qa_prompt: str,
rephrase_prompt: str,
config: dict,
callbacks=None,
):
"""
Initialize the Langchain_RAG class.
Args:
llm (LanguageModelLike): The language model instance.
memory (BaseChatMessageHistory): The chat message history instance.
retriever (BaseRetriever): The retriever instance.
qa_prompt (str): The QA prompt string.
rephrase_prompt (str): The rephrase prompt string.
"""
self.llm = llm
self.config = config
# self.memory = self.add_history_from_list(memory)
self.memory = ConversationBufferWindowMemory(
k=self.config["llm_params"]["memory_window"],
memory_key="chat_history",
return_messages=True,
output_key="answer",
max_token_limit=128,
)
self.retriever = retriever
self.qa_prompt = qa_prompt
self.rephrase_prompt = rephrase_prompt
self.store = {}
self.qa_prompt = PromptTemplate(
template=self.qa_prompt,
input_variables=["context", "chat_history", "input"],
)
self.rag_chain = CustomConversationalRetrievalChain.from_llm(
llm=llm,
chain_type="stuff",
retriever=retriever,
return_source_documents=True,
memory=self.memory,
combine_docs_chain_kwargs={"prompt": self.qa_prompt},
response_if_no_docs_found="No context found",
)
def add_history_from_list(self, history_list):
"""
TODO: Add messages from a list to the chat history.
"""
history = []
return history
async def invoke(self, user_query, config):
"""
Invoke the chain.
Args:
kwargs: The input variables.
Returns:
dict: The output variables.
"""
res = await self.rag_chain.acall(user_query["input"])
return res
class QuestionGenerator:
"""
Generate a question from the LLMs response and users input and past conversations.
"""
def __init__(self):
pass
def generate_questions(self, query, response, chat_history, context, config):
questions = return_questions(query, response, chat_history, context, config)
return questions
class Langchain_RAG_V2(BaseRAG):
def __init__(
self,
llm,
memory,
retriever,
qa_prompt: str,
rephrase_prompt: str,
config: dict,
callbacks=None,
):
"""
Initialize the Langchain_RAG class.
Args:
llm (LanguageModelLike): The language model instance.
memory (BaseChatMessageHistory): The chat message history instance.
retriever (BaseRetriever): The retriever instance.
qa_prompt (str): The QA prompt string.
rephrase_prompt (str): The rephrase prompt string.
"""
self.llm = llm
self.memory = self.add_history_from_list(memory)
self.retriever = retriever
self.qa_prompt = qa_prompt
self.rephrase_prompt = rephrase_prompt
self.store = {}
# Contextualize question prompt
contextualize_q_system_prompt = rephrase_prompt or (
"Given a chat history and the latest user question "
"which might reference context in the chat history, "
"formulate a standalone question which can be understood "
"without the chat history. Do NOT answer the question, just "
"reformulate it if needed and otherwise return it as is."
)
self.contextualize_q_prompt = ChatPromptTemplate.from_template(
contextualize_q_system_prompt
)
# History-aware retriever
self.history_aware_retriever = create_history_aware_retriever(
self.llm, self.retriever, self.contextualize_q_prompt
)
# Answer question prompt
qa_system_prompt = qa_prompt or (
"You are an assistant for question-answering tasks. Use "
"the following pieces of retrieved context to answer the "
"question. If you don't know the answer, just say that you "
"don't know. Use three sentences maximum and keep the answer "
"concise."
"\n\n"
"{context}"
)
self.qa_prompt_template = ChatPromptTemplate.from_template(qa_system_prompt)
# Question-answer chain
self.question_answer_chain = create_stuff_documents_chain(
self.llm, self.qa_prompt_template
)
# Final retrieval chain
self.rag_chain = create_retrieval_chain(
self.history_aware_retriever, self.question_answer_chain
)
self.rag_chain = CustomRunnableWithHistory(
self.rag_chain,
get_session_history=self.get_session_history,
input_messages_key="input",
history_messages_key="chat_history",
output_messages_key="answer",
history_factory_config=[
ConfigurableFieldSpec(
id="user_id",
annotation=str,
name="User ID",
description="Unique identifier for the user.",
default="",
is_shared=True,
),
ConfigurableFieldSpec(
id="conversation_id",
annotation=str,
name="Conversation ID",
description="Unique identifier for the conversation.",
default="",
is_shared=True,
),
ConfigurableFieldSpec(
id="memory_window",
annotation=int,
name="Number of Conversations",
description="Number of conversations to consider for context.",
default=1,
is_shared=True,
),
],
).with_config(run_name="Langchain_RAG_V2")
if callbacks is not None:
self.rag_chain = self.rag_chain.with_config(callbacks=callbacks)
def get_session_history(
self, user_id: str, conversation_id: str, memory_window: int
) -> BaseChatMessageHistory:
"""
Get the session history for a user and conversation.
Args:
user_id (str): The user identifier.
conversation_id (str): The conversation identifier.
memory_window (int): The number of conversations to consider for context.
Returns:
BaseChatMessageHistory: The chat message history.
"""
if (user_id, conversation_id) not in self.store:
self.store[(user_id, conversation_id)] = InMemoryHistory()
self.store[(user_id, conversation_id)].add_messages(
self.memory.messages
) # add previous messages to the store. Note: the store is in-memory.
return self.store[(user_id, conversation_id)]
async def invoke(self, user_query, config, **kwargs):
"""
Invoke the chain.
Args:
kwargs: The input variables.
Returns:
dict: The output variables.
"""
res = await self.rag_chain.ainvoke(user_query, config, **kwargs)
res["rephrase_prompt"] = self.rephrase_prompt
res["qa_prompt"] = self.qa_prompt
return res
def stream(self, user_query, config):
res = self.rag_chain.stream(user_query, config)
return res
def add_history_from_list(self, conversation_list):
"""
Add messages from a list to the chat history.
Args:
messages (list): The list of messages to add.
"""
history = ChatMessageHistory()
for idx, message in enumerate(conversation_list):
message_type = (
message.get("type", None)
if isinstance(message, dict)
else getattr(message, "type", None)
)
message_content = (
message.get("content", None)
if isinstance(message, dict)
else getattr(message, "content", None)
)
if message_type in ["human", "user_message"]:
history.add_user_message(message_content)
elif message_type in ["ai", "ai_message"]:
history.add_ai_message(message_content)
return history
|