Upload 4 files

README.md

@@ -1,4 +1,4 @@
----
+
 title: Amharic Srh Chatbot
 emoji: 👀
 colorFrom: purple
@@ -7,6 +7,3 @@ sdk: streamlit
 sdk_version: 1.44.1
 app_file: app.py
 pinned: false
----
-
-Check out the configuration reference at https://huggingface.co/docs/hub/spaces-config-reference

app.py

@@ -1,16 +1,15 @@
+
 import streamlit as st
 from chatbot_utils import AmharicChatbot
 
 st.set_page_config(page_title="Amharic SRH Chatbot", layout="centered")
 
-# Load the chatbot
 @st.cache_resource
 def load_bot():
     return AmharicChatbot("amharic_srh_qa.csv")
 
 bot = load_bot()
 
-# Inject Custom CSS
 st.markdown("""
     <style>
         .chat-container {
@@ -48,26 +47,6 @@ st.markdown("""
             align-self: flex-start;
             margin-right: auto;
         }
-        .input-area {
-            display: flex;
-            gap: 10px;
-        }
-        .input-text {
-            flex: 1;
-            padding: 10px;
-            border-radius: 5px;
-            border: 1px solid #ccc;
-        }
-        .send-btn {
-            padding: 10px 20px;
-            background-color: #4CAF50;
-            color: white;
-            border: none;
-            border-radius: 5px;
-        }
-        .send-btn:hover {
-            background-color: #45a049;
-        }
     </style>
 """, unsafe_allow_html=True)
 
@@ -76,7 +55,6 @@ st.markdown("## 🤖 አማርኛ ጤና ቻትቦት", unsafe_allow_html=True)
 st.markdown("ስለ ወሊድና የአባላዘር በሽታ ጥያቄ አሎት? እባክዎ ያቀርቡ።", unsafe_allow_html=True)
 st.markdown('<div class="chat-box" id="chat-box">', unsafe_allow_html=True)
 
-# Use session state to track messages
 if "messages" not in st.session_state:
     st.session_state.messages = []
 
@@ -84,9 +62,8 @@ for msg in st.session_state.messages:
     css_class = "user-message" if msg["sender"] == "user" else "bot-message"
     st.markdown(f'<div class="message {css_class}">{msg["text"]}</div>', unsafe_allow_html=True)
 
-st.markdown('</div>', unsafe_allow_html=True)  # Close chat-box
+st.markdown('</div>', unsafe_allow_html=True)
 
-# Input area
 with st.form(key="chat_form"):
     user_input = st.text_input("💬 ጥያቄዎን ያስገቡ:", key="input")
     submit = st.form_submit_button("መልስ አውጣ")
@@ -94,10 +71,8 @@ with st.form(key="chat_form"):
     if submit and user_input:
         st.session_state.messages.append({"sender": "user", "text": user_input})
         response = bot.get_answer(user_input)
-
         if response == "__OUT_OF_SCOPE__":
             response = "ይቅርታ፣ ይህንን ጥያቄ ማስተዋል አልቻልኩም። እባክዎ በሌላ መንገድ ይሞክሩ።"
-
         st.session_state.messages.append({"sender": "bot", "text": response})
 
-st.markdown('</div>', unsafe_allow_html=True)  # Close chat-container
+st.markdown('</div>', unsafe_allow_html=True)

chatbot_utils.py

@@ -1,3 +1,4 @@
+
 import pandas as pd
 import faiss
 from sentence_transformers import SentenceTransformer
@@ -12,7 +13,6 @@ class AmharicChatbot:
         self.build_index()
 
     def build_index(self):
-        # Encode questions using the E5 small model; "passage:" prefix for context
         self.embeddings = self.model.encode(
             ["passage: " + q for q in self.df["question"].tolist()],
             show_progress_bar=True
@@ -21,29 +21,12 @@ class AmharicChatbot:
         self.index.add(self.embeddings)
 
     def get_answer(self, user_question, k=1):
-        # Encode the user question with "query:" prefix for best retrieval
         user_embedding = self.model.encode([f"query: {user_question}"])[0].astype("float32")
         D, I = self.index.search(np.array([user_embedding]), k)
-
         top_idx = I[0][0]
         top_question = self.df.iloc[top_idx]["question"]
         top_embedding = self.model.encode([f"passage: {top_question}"])[0]
-        
-        # Cosine similarity score between user and top retrieved question
         score = cosine_similarity([user_embedding], [top_embedding])[0][0]
-
         if score < self.threshold:
             return "__OUT_OF_SCOPE__"
-        
         return self.df.iloc[top_idx]["answer"]
-
-    # Optional: retrieve top-K relevant Q&A pairs for further use
-    def get_top_k(self, user_question, k=3):
-        user_embedding = self.model.encode([f"query: {user_question}"])[0].astype("float32")
-        D, I = self.index.search(np.array([user_embedding]), k)
-        results = []
-        for idx in I[0]:
-            question = self.df.iloc[idx]["question"]
-            answer = self.df.iloc[idx]["answer"]
-            results.append((question, answer))
-        return results

requirements.txt

@@ -1,6 +1,7 @@
+
 streamlit
 sentence-transformers
 faiss-cpu
 pandas
+scikit-learn
 huggingface-hub
-scikit-learn