Update app.py
Browse files
app.py
CHANGED
@@ -23,8 +23,6 @@ import requests
|
|
23 |
import json
|
24 |
import os
|
25 |
|
26 |
-
|
27 |
-
|
28 |
# set this key as an environment variable
|
29 |
os.environ["HUGGINGFACEHUB_API_TOKEN"] = st.secrets['huggingface_token']
|
30 |
|
@@ -36,29 +34,27 @@ Public_Url = 'https://jikoni-tmodel.hf.space' #endpoint
|
|
36 |
|
37 |
def translate(userinput, target_lang, source_lang=None):
|
38 |
if source_lang:
|
39 |
-
|
40 |
-
|
41 |
-
|
42 |
-
|
43 |
-
|
44 |
}
|
45 |
-
|
46 |
-
|
47 |
-
|
48 |
-
|
49 |
-
|
50 |
-
|
51 |
else:
|
52 |
-
|
53 |
-
|
54 |
-
|
55 |
-
|
56 |
-
|
57 |
-
|
58 |
-
|
59 |
-
|
60 |
-
|
61 |
-
translation = result['translated_text']
|
62 |
return source_lange, translation
|
63 |
|
64 |
def get_pdf_text(pdf : Union[str, bytes, bytearray]) -> str:
|
@@ -70,7 +66,6 @@ def get_pdf_text(pdf : Union[str, bytes, bytearray]) -> str:
|
|
70 |
pdf_text += text
|
71 |
return text
|
72 |
|
73 |
-
|
74 |
def get_text_chunks(text:str) ->list:
|
75 |
text_splitter = CharacterTextSplitter(
|
76 |
separator="\n", chunk_size=1500, chunk_overlap=300, length_function=len
|
@@ -78,7 +73,6 @@ def get_text_chunks(text:str) ->list:
|
|
78 |
chunks = text_splitter.split_text(text)
|
79 |
return chunks
|
80 |
|
81 |
-
|
82 |
def get_vectorstore(text_chunks : list) -> FAISS:
|
83 |
model = "sentence-transformers/paraphrase-multilingual-MiniLM-L12-v2"
|
84 |
encode_kwargs = {
|
@@ -90,13 +84,11 @@ def get_vectorstore(text_chunks : list) -> FAISS:
|
|
90 |
vectorstore = FAISS.from_texts(texts=text_chunks, embedding=embeddings)
|
91 |
return vectorstore
|
92 |
|
93 |
-
|
94 |
def get_conversation_chain(vectorstore:FAISS) -> ConversationalRetrievalChain:
|
95 |
llm = HuggingFaceHub(
|
96 |
-
#repo_id="mistralai/Mixtral-8x7B-Instruct-v0.1",
|
97 |
repo_id="TheBloke/Mixtral-8x7B-Instruct-v0.1-GGUF",
|
98 |
task="text-generation",
|
99 |
-
model_kwargs={"temperature": 0.5, "max_length": 1048}
|
100 |
)
|
101 |
|
102 |
memory = ConversationBufferMemory(memory_key="chat_history", return_messages=True)
|
@@ -105,7 +97,6 @@ def get_conversation_chain(vectorstore:FAISS) -> ConversationalRetrievalChain:
|
|
105 |
)
|
106 |
return conversation_chain
|
107 |
|
108 |
-
|
109 |
st.markdown ("""
|
110 |
<style> div.stSpinner > div {
|
111 |
text-align:center;
|
@@ -151,7 +142,6 @@ def main():
|
|
151 |
st.session_state.conversation = get_conversation_chain(vectorstore)
|
152 |
st.info("done")
|
153 |
|
154 |
-
#user_question = st.text_input("chat with your pdf ...")
|
155 |
# show user input
|
156 |
if "messages" not in st.session_state:
|
157 |
st.session_state.messages = []
|
@@ -179,6 +169,5 @@ def main():
|
|
179 |
# Signature
|
180 |
st.markdown(footer, unsafe_allow_html=True)
|
181 |
|
182 |
-
|
183 |
if __name__ == '__main__':
|
184 |
main()
|
|
|
23 |
import json
|
24 |
import os
|
25 |
|
|
|
|
|
26 |
# set this key as an environment variable
|
27 |
os.environ["HUGGINGFACEHUB_API_TOKEN"] = st.secrets['huggingface_token']
|
28 |
|
|
|
34 |
|
35 |
def translate(userinput, target_lang, source_lang=None):
|
36 |
if source_lang:
|
37 |
+
url = f"{Public_Url}/translate_enter/"
|
38 |
+
data = {
|
39 |
+
"userinput": userinput,
|
40 |
+
"source_lang": source_lang,
|
41 |
+
"target_lang": target_lang,
|
42 |
}
|
43 |
+
response = requests.post(url, json=data)
|
44 |
+
result = response.json()
|
45 |
+
print(type(result))
|
46 |
+
source_lange = source_lang
|
47 |
+
translation = result['translated_text']
|
|
|
48 |
else:
|
49 |
+
url = f"{Public_Url}/translate_detect/"
|
50 |
+
data = {
|
51 |
+
"userinput": userinput,
|
52 |
+
"target_lang": target_lang,
|
53 |
+
}
|
54 |
+
response = requests.post(url, json=data)
|
55 |
+
result = response.json()
|
56 |
+
source_lange = result['source_language']
|
57 |
+
translation = result['translated_text']
|
|
|
58 |
return source_lange, translation
|
59 |
|
60 |
def get_pdf_text(pdf : Union[str, bytes, bytearray]) -> str:
|
|
|
66 |
pdf_text += text
|
67 |
return text
|
68 |
|
|
|
69 |
def get_text_chunks(text:str) ->list:
|
70 |
text_splitter = CharacterTextSplitter(
|
71 |
separator="\n", chunk_size=1500, chunk_overlap=300, length_function=len
|
|
|
73 |
chunks = text_splitter.split_text(text)
|
74 |
return chunks
|
75 |
|
|
|
76 |
def get_vectorstore(text_chunks : list) -> FAISS:
|
77 |
model = "sentence-transformers/paraphrase-multilingual-MiniLM-L12-v2"
|
78 |
encode_kwargs = {
|
|
|
84 |
vectorstore = FAISS.from_texts(texts=text_chunks, embedding=embeddings)
|
85 |
return vectorstore
|
86 |
|
|
|
87 |
def get_conversation_chain(vectorstore:FAISS) -> ConversationalRetrievalChain:
|
88 |
llm = HuggingFaceHub(
|
|
|
89 |
repo_id="TheBloke/Mixtral-8x7B-Instruct-v0.1-GGUF",
|
90 |
task="text-generation",
|
91 |
+
model_kwargs={"temperature": 0.5, "max_length": 1048}
|
92 |
)
|
93 |
|
94 |
memory = ConversationBufferMemory(memory_key="chat_history", return_messages=True)
|
|
|
97 |
)
|
98 |
return conversation_chain
|
99 |
|
|
|
100 |
st.markdown ("""
|
101 |
<style> div.stSpinner > div {
|
102 |
text-align:center;
|
|
|
142 |
st.session_state.conversation = get_conversation_chain(vectorstore)
|
143 |
st.info("done")
|
144 |
|
|
|
145 |
# show user input
|
146 |
if "messages" not in st.session_state:
|
147 |
st.session_state.messages = []
|
|
|
169 |
# Signature
|
170 |
st.markdown(footer, unsafe_allow_html=True)
|
171 |
|
|
|
172 |
if __name__ == '__main__':
|
173 |
main()
|