Spaces:
Sleeping
Sleeping
Commit
·
8728fc3
1
Parent(s):
64cbc9f
Update Streamlit app
Browse files- .env +1 -0
- app.py +74 -2
- rag_notebook.ipynb +0 -2
.env
ADDED
@@ -0,0 +1 @@
|
|
|
|
|
1 |
+
COHERE_API_KEY=I2LwLsW10InzTzGZ0WMriKxHUHq78E5pSVkl3MTe
|
app.py
CHANGED
@@ -1,4 +1,76 @@
|
|
1 |
import streamlit as st
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2 |
|
3 |
-
|
4 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
import streamlit as st
|
2 |
+
import bs4
|
3 |
+
from langchain_community.document_loaders import WebBaseLoader
|
4 |
+
from langchain_text_splitters import RecursiveCharacterTextSplitter
|
5 |
+
from langchain.vectorstores import FAISS
|
6 |
+
from langchain.embeddings import HuggingFaceEmbeddings
|
7 |
+
from langchain_cohere import ChatCohere
|
8 |
+
from langchain_core.output_parsers import StrOutputParser
|
9 |
+
from langchain_core.runnables import RunnablePassthrough
|
10 |
+
from langchain_core.prompts import PromptTemplate
|
11 |
+
from dotenv import load_dotenv
|
12 |
+
load_dotenv('.env')
|
13 |
|
14 |
+
st.header("MKOM UGM RAG App")
|
15 |
+
|
16 |
+
# Only keep post title, headers, and content from the full HTML.
|
17 |
+
bs4_strainer = bs4.SoupStrainer(class_=("post-title", "post-header", "post-content"))
|
18 |
+
loader = WebBaseLoader(
|
19 |
+
web_paths=(
|
20 |
+
"https://um.ugm.ac.id/ragam-seleksi-pascasarjana/",
|
21 |
+
"https://um.ugm.ac.id/persyaratan-pendaftaran-magister/",
|
22 |
+
"https://um.ugm.ac.id/persyaratan-pendaftaran-program-spesialis/",
|
23 |
+
"https://um.ugm.ac.id/persyaratan-pendaftaran-subspesialis/",
|
24 |
+
"https://um.ugm.ac.id/persyaratan-pendaftaran-doktor/",
|
25 |
+
"https://um.ugm.ac.id/prosedur-pendaftaran-magister/",
|
26 |
+
"https://um.ugm.ac.id/prosedur-pendaftaran-program-spesialis/",
|
27 |
+
"https://um.ugm.ac.id/prosedur-pendaftaran-program-subspesialis/",
|
28 |
+
"https://um.ugm.ac.id/prosedur-pendaftaran-doktor-2/",
|
29 |
+
"https://um.ugm.ac.id/program-studi-program-magister-2/",
|
30 |
+
"https://um.ugm.ac.id/program-studi-dan-daya-tampung-program-spesialis/",
|
31 |
+
"https://um.ugm.ac.id/program-studi-program-doktor/",
|
32 |
+
"https://um.ugm.ac.id/jadwal-seleksi-magister-dan-doktor/",
|
33 |
+
"https://um.ugm.ac.id/jadwal-kegiatan-seleksi-program-spesialis/",
|
34 |
+
"https://mkom.ugm.ac.id/alur-pendaftaran-magister/",
|
35 |
+
"https://mkom.ugm.ac.id/informasi-pendaftaran-program-pra-s2-ilmu-komputer/",
|
36 |
+
"https://mkom.ugm.ac.id/informasi-pendaftaran-program-s2-magister/",
|
37 |
+
"https://mkom.ugm.ac.id/program-dual-degree-double-degree-magister-ilmu-komputer/",
|
38 |
+
"https://mkom.ugm.ac.id/informasi-pendaftaran-program-s3-doktor/"
|
39 |
+
),
|
40 |
+
bs_kwargs={"parse_only": bs4_strainer},
|
41 |
+
)
|
42 |
+
docs = loader.load()
|
43 |
+
text_splitter = RecursiveCharacterTextSplitter(
|
44 |
+
chunk_size=1000, chunk_overlap=200, add_start_index=True
|
45 |
+
)
|
46 |
+
all_splits = text_splitter.split_documents(docs)
|
47 |
+
vectorstore = FAISS.from_documents(all_splits, HuggingFaceEmbeddings(model_name='firqaaa/indo-sentence-bert-base'))
|
48 |
+
retriever = vectorstore.as_retriever(search_type="similarity", search_kwargs={"k": 6})
|
49 |
+
llm = ChatCohere(model="command-r")
|
50 |
+
|
51 |
+
def format_docs(docs):
|
52 |
+
return "\n\n".join(doc.page_content for doc in docs)
|
53 |
+
|
54 |
+
template = """Gunakan konteks berikut untuk menjawab pertanyaan pada bagian akhir.
|
55 |
+
Jika kamu tidak tahu jawabannya, katakan saja bahwa kamu tidak tahu, jangan mencoba untuk mengarang jawaban.
|
56 |
+
Selalu katakan "Terima kasih sudah bertanya!" pada setiap akhir jawaban.
|
57 |
+
|
58 |
+
{context}
|
59 |
+
|
60 |
+
Pertanyaan: {question}
|
61 |
+
|
62 |
+
Jawaban:"""
|
63 |
+
|
64 |
+
custom_rag_prompt = PromptTemplate.from_template(template)
|
65 |
+
|
66 |
+
rag_chain = (
|
67 |
+
{"context": retriever | format_docs, "question": RunnablePassthrough()}
|
68 |
+
| custom_rag_prompt
|
69 |
+
| llm
|
70 |
+
| StrOutputParser()
|
71 |
+
)
|
72 |
+
|
73 |
+
question = st.text_input("Tanya ujian masuk Pascasarjana Universitas Gadjah Mada")
|
74 |
+
if question:
|
75 |
+
response = rag_chain.invoke(question)
|
76 |
+
st.write(response)
|
rag_notebook.ipynb
CHANGED
@@ -162,11 +162,9 @@
|
|
162 |
}
|
163 |
],
|
164 |
"source": [
|
165 |
-
"from langchain_chroma import Chroma\n",
|
166 |
"from langchain.vectorstores import FAISS\n",
|
167 |
"from langchain.embeddings import HuggingFaceEmbeddings\n",
|
168 |
"\n",
|
169 |
-
"# vectorstore = Chroma.from_documents(documents=all_splits, embedding=HuggingFaceEmbeddings(model_name='firqaaa/indo-sentence-bert-base'))\n",
|
170 |
"vectorstore = FAISS.from_documents(all_splits, HuggingFaceEmbeddings(model_name='firqaaa/indo-sentence-bert-base'))"
|
171 |
]
|
172 |
},
|
|
|
162 |
}
|
163 |
],
|
164 |
"source": [
|
|
|
165 |
"from langchain.vectorstores import FAISS\n",
|
166 |
"from langchain.embeddings import HuggingFaceEmbeddings\n",
|
167 |
"\n",
|
|
|
168 |
"vectorstore = FAISS.from_documents(all_splits, HuggingFaceEmbeddings(model_name='firqaaa/indo-sentence-bert-base'))"
|
169 |
]
|
170 |
},
|