Spaces:
Sleeping
Sleeping
Update app.py
Browse files
app.py
CHANGED
@@ -4,19 +4,19 @@ import torchaudio
|
|
4 |
import soundfile as sf
|
5 |
from pathlib import Path
|
6 |
from transformers import pipeline, AutoTokenizer
|
|
|
|
|
7 |
from langchain.memory import ConversationBufferMemory
|
8 |
from langchain.chains import ConversationalRetrievalChain
|
9 |
-
from langchain_community.llms import HuggingFaceHub
|
10 |
-
from langchain_community.embeddings import HuggingFaceEmbeddings
|
11 |
from langchain_community.vectorstores import FAISS
|
12 |
-
from
|
13 |
import os
|
14 |
from dotenv import load_dotenv
|
15 |
|
16 |
# Load environment variables
|
17 |
load_dotenv()
|
18 |
|
19 |
-
# CSS
|
20 |
css = """
|
21 |
<style>
|
22 |
.chat-message { padding: 1.5rem; border-radius: 0.5rem; margin-bottom: 1rem; display: flex; }
|
@@ -27,17 +27,20 @@ css = """
|
|
27 |
</style>
|
28 |
"""
|
29 |
|
30 |
-
#
|
31 |
PROMPT_TEMPLATE = """
|
32 |
You are a professional therapist who speaks Moroccan Arabic (Darija).
|
33 |
Respond with empathy and use therapeutic techniques.
|
34 |
Always respond in Darija unless specifically asked to use another language.
|
35 |
|
36 |
-
|
37 |
-
|
38 |
-
Current Question: {question}
|
39 |
|
40 |
-
|
|
|
|
|
|
|
|
|
41 |
"""
|
42 |
|
43 |
class DarijaTherapist:
|
@@ -48,8 +51,8 @@ class DarijaTherapist:
|
|
48 |
|
49 |
def setup_models(self):
|
50 |
try:
|
|
|
51 |
tokenizer = AutoTokenizer.from_pretrained("facebook/seamless-m4t-v2-large")
|
52 |
-
|
53 |
self.device = "cuda" if torch.cuda.is_available() else "cpu"
|
54 |
self.asr_pipe = pipeline(
|
55 |
"automatic-speech-recognition",
|
@@ -58,17 +61,25 @@ class DarijaTherapist:
|
|
58 |
device=self.device
|
59 |
)
|
60 |
|
61 |
-
|
62 |
-
|
63 |
-
|
64 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
65 |
huggingfacehub_api_token=os.getenv("HUGGINGFACE_API_TOKEN")
|
66 |
)
|
67 |
|
68 |
-
|
69 |
-
|
|
|
70 |
)
|
71 |
|
|
|
72 |
self.vectorstore = FAISS.from_texts(
|
73 |
["Initial therapeutic context"],
|
74 |
self.embeddings
|
@@ -76,14 +87,13 @@ class DarijaTherapist:
|
|
76 |
except Exception as e:
|
77 |
st.error(f"Error setting up models: {str(e)}")
|
78 |
st.stop()
|
79 |
-
|
80 |
def setup_memory(self):
|
81 |
self.memory = ConversationBufferMemory(
|
82 |
memory_key="chat_history",
|
83 |
return_messages=True
|
84 |
)
|
85 |
|
86 |
-
# Updated chain creation with correct prompt
|
87 |
qa_prompt = PromptTemplate(
|
88 |
template=PROMPT_TEMPLATE,
|
89 |
input_variables=["context", "chat_history", "question"]
|
@@ -94,11 +104,9 @@ class DarijaTherapist:
|
|
94 |
retriever=self.vectorstore.as_retriever(),
|
95 |
memory=self.memory,
|
96 |
combine_docs_chain_kwargs={"prompt": qa_prompt},
|
97 |
-
return_source_documents=True
|
98 |
-
chain_type="stuff"
|
99 |
)
|
100 |
|
101 |
-
# Rest of the methods remain the same
|
102 |
def initialize_session_state(self):
|
103 |
if "messages" not in st.session_state:
|
104 |
st.session_state.messages = []
|
@@ -138,7 +146,10 @@ class DarijaTherapist:
|
|
138 |
|
139 |
def get_ai_response(self, user_input):
|
140 |
try:
|
141 |
-
response = self.conversation_chain({
|
|
|
|
|
|
|
142 |
return response['answer']
|
143 |
except Exception as e:
|
144 |
st.error(f"Error getting AI response: {str(e)}")
|
|
|
4 |
import soundfile as sf
|
5 |
from pathlib import Path
|
6 |
from transformers import pipeline, AutoTokenizer
|
7 |
+
from langchain_community.llms import HuggingFaceEndpoint
|
8 |
+
from langchain_community.embeddings import HuggingFaceBgeEmbeddings
|
9 |
from langchain.memory import ConversationBufferMemory
|
10 |
from langchain.chains import ConversationalRetrievalChain
|
|
|
|
|
11 |
from langchain_community.vectorstores import FAISS
|
12 |
+
from langchain.prompts import PromptTemplate
|
13 |
import os
|
14 |
from dotenv import load_dotenv
|
15 |
|
16 |
# Load environment variables
|
17 |
load_dotenv()
|
18 |
|
19 |
+
# CSS styling
|
20 |
css = """
|
21 |
<style>
|
22 |
.chat-message { padding: 1.5rem; border-radius: 0.5rem; margin-bottom: 1rem; display: flex; }
|
|
|
27 |
</style>
|
28 |
"""
|
29 |
|
30 |
+
# Prompt template
|
31 |
PROMPT_TEMPLATE = """
|
32 |
You are a professional therapist who speaks Moroccan Arabic (Darija).
|
33 |
Respond with empathy and use therapeutic techniques.
|
34 |
Always respond in Darija unless specifically asked to use another language.
|
35 |
|
36 |
+
Previous conversation:
|
37 |
+
{chat_history}
|
|
|
38 |
|
39 |
+
User message: {question}
|
40 |
+
|
41 |
+
Additional context: {context}
|
42 |
+
|
43 |
+
Therapeutic response in Darija:
|
44 |
"""
|
45 |
|
46 |
class DarijaTherapist:
|
|
|
51 |
|
52 |
def setup_models(self):
|
53 |
try:
|
54 |
+
# Speech recognition setup
|
55 |
tokenizer = AutoTokenizer.from_pretrained("facebook/seamless-m4t-v2-large")
|
|
|
56 |
self.device = "cuda" if torch.cuda.is_available() else "cpu"
|
57 |
self.asr_pipe = pipeline(
|
58 |
"automatic-speech-recognition",
|
|
|
61 |
device=self.device
|
62 |
)
|
63 |
|
64 |
+
# LLM setup
|
65 |
+
self.llm = HuggingFaceEndpoint(
|
66 |
+
endpoint_url="https://api-inference.huggingface.co/models/MBZUAI-Paris/Atlas-Chat-9B",
|
67 |
+
task="text-generation",
|
68 |
+
model_kwargs={
|
69 |
+
"temperature": 0.7,
|
70 |
+
"max_length": 512,
|
71 |
+
"do_sample": True,
|
72 |
+
"return_full_text": False
|
73 |
+
},
|
74 |
huggingfacehub_api_token=os.getenv("HUGGINGFACE_API_TOKEN")
|
75 |
)
|
76 |
|
77 |
+
# Embeddings setup
|
78 |
+
self.embeddings = HuggingFaceBgeEmbeddings(
|
79 |
+
model_name="BAAI/bge-large-en"
|
80 |
)
|
81 |
|
82 |
+
# Vector store setup
|
83 |
self.vectorstore = FAISS.from_texts(
|
84 |
["Initial therapeutic context"],
|
85 |
self.embeddings
|
|
|
87 |
except Exception as e:
|
88 |
st.error(f"Error setting up models: {str(e)}")
|
89 |
st.stop()
|
90 |
+
|
91 |
def setup_memory(self):
|
92 |
self.memory = ConversationBufferMemory(
|
93 |
memory_key="chat_history",
|
94 |
return_messages=True
|
95 |
)
|
96 |
|
|
|
97 |
qa_prompt = PromptTemplate(
|
98 |
template=PROMPT_TEMPLATE,
|
99 |
input_variables=["context", "chat_history", "question"]
|
|
|
104 |
retriever=self.vectorstore.as_retriever(),
|
105 |
memory=self.memory,
|
106 |
combine_docs_chain_kwargs={"prompt": qa_prompt},
|
107 |
+
return_source_documents=True
|
|
|
108 |
)
|
109 |
|
|
|
110 |
def initialize_session_state(self):
|
111 |
if "messages" not in st.session_state:
|
112 |
st.session_state.messages = []
|
|
|
146 |
|
147 |
def get_ai_response(self, user_input):
|
148 |
try:
|
149 |
+
response = self.conversation_chain({
|
150 |
+
"question": user_input,
|
151 |
+
"chat_history": self.memory.chat_memory.messages
|
152 |
+
})
|
153 |
return response['answer']
|
154 |
except Exception as e:
|
155 |
st.error(f"Error getting AI response: {str(e)}")
|