Update scripts/router_chain.py
Browse files- scripts/router_chain.py +38 -13
scripts/router_chain.py
CHANGED
@@ -1,32 +1,57 @@
|
|
1 |
from typing import Dict, Any
|
2 |
from langchain_openai import ChatOpenAI
|
3 |
from langchain.prompts import ChatPromptTemplate
|
|
|
4 |
from scripts.rag_chat import build_general_qa_chain
|
5 |
|
6 |
def build_router_chain(model_name=None):
|
7 |
general_qa = build_general_qa_chain(model_name=model_name)
|
8 |
llm = ChatOpenAI(model_name=model_name or "gpt-4o-mini", temperature=0.0)
|
9 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
10 |
class Router:
|
11 |
def invoke(self, input_dict: Dict[str, Any]):
|
12 |
-
|
13 |
-
|
|
|
|
|
|
|
14 |
prompt = ChatPromptTemplate.from_template(
|
15 |
"As a coding assistant, help with this Python question.\nQuestion: {input}\nAnswer:"
|
16 |
)
|
17 |
-
chain = prompt | llm
|
18 |
-
return {"result": chain.invoke({"input": input_dict["input"]})
|
19 |
-
|
|
|
20 |
prompt = ChatPromptTemplate.from_template(
|
21 |
"Provide a concise summary about: {input}\nSummary:"
|
22 |
)
|
23 |
-
chain = prompt | llm
|
24 |
-
return {"result": chain.invoke({"input": input_dict["input"]})
|
25 |
-
|
26 |
-
|
27 |
-
|
28 |
-
|
29 |
-
|
30 |
-
|
|
|
|
|
|
|
|
|
31 |
|
32 |
return Router()
|
|
|
1 |
from typing import Dict, Any
|
2 |
from langchain_openai import ChatOpenAI
|
3 |
from langchain.prompts import ChatPromptTemplate
|
4 |
+
from langchain.schema import StrOutputParser
|
5 |
from scripts.rag_chat import build_general_qa_chain
|
6 |
|
7 |
def build_router_chain(model_name=None):
|
8 |
general_qa = build_general_qa_chain(model_name=model_name)
|
9 |
llm = ChatOpenAI(model_name=model_name or "gpt-4o-mini", temperature=0.0)
|
10 |
|
11 |
+
# This prompt asks the LLM to choose which "mode" to use
|
12 |
+
router_prompt = ChatPromptTemplate.from_template("""
|
13 |
+
You are a routing assistant for a chatbot.
|
14 |
+
Classify the following user request into one of these categories:
|
15 |
+
- "code" for programming or debugging
|
16 |
+
- "summarize" for summary requests
|
17 |
+
- "calculate" for math or numeric calculations
|
18 |
+
- "general" for general Q&A using course files
|
19 |
+
|
20 |
+
Return ONLY the category word.
|
21 |
+
|
22 |
+
User request: {input}
|
23 |
+
""")
|
24 |
+
|
25 |
+
router_chain = router_prompt | llm | StrOutputParser()
|
26 |
+
|
27 |
class Router:
|
28 |
def invoke(self, input_dict: Dict[str, Any]):
|
29 |
+
category = router_chain.invoke({"input": input_dict["input"]}).strip().lower()
|
30 |
+
|
31 |
+
print(f"[ROUTER] User query routed to category: {category}")
|
32 |
+
|
33 |
+
if category == "code":
|
34 |
prompt = ChatPromptTemplate.from_template(
|
35 |
"As a coding assistant, help with this Python question.\nQuestion: {input}\nAnswer:"
|
36 |
)
|
37 |
+
chain = prompt | llm | StrOutputParser()
|
38 |
+
return {"result": chain.invoke({"input": input_dict["input"]})}
|
39 |
+
|
40 |
+
elif category == "summarize":
|
41 |
prompt = ChatPromptTemplate.from_template(
|
42 |
"Provide a concise summary about: {input}\nSummary:"
|
43 |
)
|
44 |
+
chain = prompt | llm | StrOutputParser()
|
45 |
+
return {"result": chain.invoke({"input": input_dict["input"]})}
|
46 |
+
|
47 |
+
elif category == "calculate":
|
48 |
+
prompt = ChatPromptTemplate.from_template(
|
49 |
+
"Solve the following calculation step-by-step:\n{input}"
|
50 |
+
)
|
51 |
+
chain = prompt | llm | StrOutputParser()
|
52 |
+
return {"result": chain.invoke({"input": input_dict["input"]})}
|
53 |
+
|
54 |
+
else: # "general"
|
55 |
+
return general_qa({"query": input_dict["input"]})
|
56 |
|
57 |
return Router()
|