Update app.py

app.py

@@ -1,4 +1,4 @@
-# app.py (robust version)
+# app.py (robust - paste entire file)
 import streamlit as st
 import torch
 import transformers
@@ -9,177 +9,108 @@ try:
 except Exception:
     from transformers.pipelines import pipeline  # fallback
 
-st.set_page_config(
-    page_title="GUVI AI Assistant",
-    page_icon="🤖",
-    layout="wide",
-    initial_sidebar_state="expanded"
-)
+st.set_page_config(page_title="GUVI AI Assistant", page_icon="🤖", layout="wide")
 
 @st.cache_resource
 def load_models():
     try:
-        translator = pipeline(
-            task="translation",
-            model="facebook/mbart-large-50-many-to-many-mmt",
-            device=0 if torch.cuda.is_available() else -1
-        )
-    except Exception as e:
-        st.error(f"Translator load failed: {e}")
-        raise
-
-    try:
-        generator = pipeline(
-            task="text-generation",
-            model="aswinprasath31/guvi-gpt2-finetuned",
-            device=0 if torch.cuda.is_available() else -1,
-            return_full_text=False
-        )
+        translator = pipeline("translation", model="facebook/mbart-large-50-many-to-many-mmt",
+                              device=0 if torch.cuda.is_available() else -1)
+        generator = pipeline("text-generation", model="aswinprasath31/guvi-gpt2-finetuned",
+                             device=0 if torch.cuda.is_available() else -1, return_full_text=False)
+        detector = pipeline("text-classification", model="papluca/xlm-roberta-base-language-detection",
+                            device=0 if torch.cuda.is_available() else -1)
     except Exception as e:
-        st.error(f"Generator load failed: {e}")
+        st.error(f"Model load failed: {e}")
         raise
-
-    try:
-        detector = pipeline(
-            task="text-classification",
-            model="papluca/xlm-roberta-base-language-detection",
-            device=0 if torch.cuda.is_available() else -1
-        )
-    except Exception as e:
-        st.error(f"Detector load failed: {e}")
-        raise
-
     return translator, generator, detector
 
 with st.spinner("Initializing AI models, please wait..."):
     translator, generator, detector = load_models()
 
 MODEL_LANG_MAP = {
-    "ar": "ar_AR", "cs": "cs_CZ", "de": "de_DE", "en": "en_XX", "es": "es_XX", "et": "et_EE", "fi": "fi_FI",
-    "fr": "fr_XX", "gu": "gu_IN", "hi": "hi_IN", "it": "it_IT", "ja": "ja_XX", "kk": "kk_KZ", "ko": "ko_KR",
-    "lt": "lt_LT", "lv": "lv_LV", "my": "my_MM", "ne": "ne_NP", "nl": "nl_XX", "ro": "ro_RO", "ru": "ru_RU",
-    "si": "si_LK", "tr": "tr_TR", "vi": "vi_VN", "zh": "zh_CN", "af": "af_ZA", "az": "az_AZ", "bn": "bn_IN",
-    "fa": "fa_IR", "he": "he_IL", "hr": "hr_HR", "id": "id_ID", "ka": "ka_GE", "km": "km_KH", "mk": "mk_MK",
-    "ml": "ml_IN", "mn": "mn_MN", "mr": "mr_IN", "pl": "pl_PL", "ps": "ps_AF", "pt": "pt_XX", "sv": "sv_SE",
-    "sw": "sw_KE", "ta": "ta_IN", "te": "te_IN", "th": "th_TH", "tl": "tl_XX", "uk": "uk_UA", "ur": "ur_PK",
-    "xh": "xh_ZA", "gl": "gl_ES", "sl": "sl_SI"
+    "ar": "ar_AR", "cs": "cs_CZ", "de": "de_DE", "en": "en_XX", "es": "es_XX", "fr": "fr_XX",
+    "gu": "gu_IN", "hi": "hi_IN", "it": "it_IT", "ja": "ja_XX", "ko": "ko_KR", "ml": "ml_IN",
+    "mr": "mr_IN", "pt": "pt_XX", "ta": "ta_IN", "te": "te_IN", "zh": "zh_CN"
 }
 
-with st.sidebar:
-    st.title("About the Chatbot")
-    st.info("Advanced multilingual AI assistant for GUVI (fine-tuned GPT-2).")
-    st.markdown("---")
-    st.header("Tech Stack")
-    st.markdown("- Streamlit\n- Hugging Face Transformers\n- PyTorch")
-
+st.sidebar.title("About the Chatbot")
+st.sidebar.info("Advanced multilingual AI assistant for GUVI (fine-tuned GPT-2).")
 st.title("GUVI AI Learning Assistant 🤖")
 
 if "messages" not in st.session_state:
-    st.session_state.messages = [{"role": "assistant", "content": "Hello! How can I help you with GUVI's courses today?"}]
-
-for message in st.session_state.messages:
-    with st.chat_message(message["role"]):
-        st.markdown(message["content"])
+    st.session_state.messages = [{"role":"assistant","content":"Hello! How can I help you with GUVI's courses today?"}]
+for msg in st.session_state.messages:
+    with st.chat_message(msg["role"]):
+        st.markdown(msg["content"])
 
 SYSTEM_PROMPT = (
-    "You are GUVI Assistant — a helpful, factual assistant for GUVI platform questions. "
-    "Answer succinctly and factually. If you are not sure, say: 'I don't know — please check GUVI's site or ask support.' "
-    "Do not include promotional claims or invented statistics."
+    "You are GUVI Assistant — helpful, factual, concise. "
+    "If unsure, reply: 'I don't know — please check GUVI's site or ask support.' "
+    "Do NOT invent statistics or promotional claims."
 )
+FEW_SHOT = ("Q: What is Data Science?\nA: Data Science is extracting insights from data "
+           "using statistics, ML, and visualization.\n---\n")
 
-FEW_SHOT_EXAMPLES = (
-    "Example:\n"
-    "Q: What is Data Science?\n"
-    "A: Data Science is the field of extracting insights from data using statistics, machine learning, and visualization.\n"
-    "---\n"
-)
-
-def build_prompt(query):
-    return f"{SYSTEM_PROMPT}\n\n{FEW_SHOT_EXAMPLES}\nUser: {query}\nAssistant:"
+def build_prompt(q): return f"{SYSTEM_PROMPT}\n\n{FEW_SHOT}\nUser: {q}\nAssistant:"
 
 def clean_output(text):
-    if not text:
-        return text
-    if "Assistant:" in text:
-        text = text.split("Assistant:")[-1]
-    lines = [ln.strip() for ln in text.splitlines() if ln.strip()]
-    if not lines:
-        return text.strip()
+    if not text: return text
+    if "Assistant:" in text: text = text.split("Assistant:")[-1]
+    lines = [l.strip() for l in text.splitlines() if l.strip()]
+    if not lines: return text.strip()
     cleaned = [lines[0]]
     for ln in lines[1:]:
         if ln != cleaned[-1]:
             cleaned.append(ln)
-    for p in ["we have taken", "for FREE", "GEEKOINS", "over 2,000"]:
+    for p in ["we have taken","GEEKOINS","for FREE","over 2,000"]:
         cleaned = [ln.replace(p, "") for ln in cleaned]
-    return "\n".join([ln for ln in cleaned if ln.strip()]).strip()
+    return "\n".join([ln for ln in cleaned if ln]).strip()
 
 def get_bot_response(user_input):
     if len(user_input.split()) < 3:
-        return "That's a broad topic! Could you ask a more specific question?"
-
+        return "Please ask a more specific question."
     try:
-        detection_result = detector(user_input)
-        detected_label = detection_result[0].get("label", "").lower()
-        lang2 = detected_label[:2] if detected_label and detected_label.isalpha() else "en"
+        det = detector(user_input)
+        label = det[0].get("label","").lower()
+        lang2 = label[:2] if label and label.isalpha() else "en"
     except Exception:
         lang2 = "en"
-
-    src_lang_code = MODEL_LANG_MAP.get(lang2, "en_XX")
-
-    if src_lang_code == "en_XX":
-        english_query = user_input
+    src = MODEL_LANG_MAP.get(lang2, "en_XX")
+    if src == "en_XX":
+        en_q = user_input
     else:
         try:
-            translated_to_en = translator(user_input, src_lang=src_lang_code, tgt_lang="en_XX")
-            english_query = translated_to_en[0].get("translation_text", user_input)
+            en_q = translator(user_input, src_lang=src, tgt_lang="en_XX")[0].get("translation_text", user_input)
         except Exception:
-            english_query = user_input
-
-    prompt = build_prompt(english_query)
-
+            en_q = user_input
+    prompt = build_prompt(en_q)
     try:
-        gen_out = generator(
-            prompt,
-            max_new_tokens=150,
-            do_sample=True,
-            temperature=0.22,
-            top_p=0.92,
-            top_k=50,
-            repetition_penalty=1.2,
-            num_return_sequences=1
-        )
-        if isinstance(gen_out, list) and isinstance(gen_out[0], dict):
-            english_response = gen_out[0].get("generated_text") or gen_out[0].get("text") or ""
+        out = generator(prompt, max_new_tokens=150, do_sample=True, temperature=0.22,
+                        top_p=0.92, top_k=50, repetition_penalty=1.2, num_return_sequences=1)
+        if isinstance(out, list) and isinstance(out[0], dict):
+            en_resp = out[0].get("generated_text") or out[0].get("text","")
         else:
-            english_response = str(gen_out[0]) if isinstance(gen_out, list) else str(gen_out)
+            en_resp = str(out[0]) if isinstance(out, list) else str(out)
     except Exception as e:
         st.error(f"Generation error: {e}")
-        return "Sorry, I'm having trouble generating a response right now."
-
-    english_response = clean_output(english_response)
-
-    if src_lang_code == "en_XX":
-        final_response = english_response
-    else:
-        try:
-            translated_to_orig = translator(english_response, src_lang="en_XX", tgt_lang=src_lang_code)
-            final_response = translated_to_orig[0].get("translation_text", english_response)
-        except Exception:
-            final_response = english_response
-
-    return final_response
+        return "Sorry, generation failed."
+    en_resp = clean_output(en_resp)
+    if src == "en_XX":
+        return en_resp
+    try:
+        return translator(en_resp, src_lang="en_XX", tgt_lang=src)[0].get("translation_text", en_resp)
+    except Exception:
+        return en_resp
 
 if prompt := st.chat_input("Ask me about Data Science, Python, or other courses..."):
-    st.session_state.messages.append({"role": "user", "content": prompt})
-    with st.chat_message("user"):
-        st.markdown(prompt)
-
+    st.session_state.messages.append({"role":"user","content":prompt})
+    with st.chat_message("user"): st.markdown(prompt)
     with st.chat_message("assistant"):
         with st.spinner("Thinking..."):
-            response = get_bot_response(prompt)
-        st.markdown(response)
-
-    st.session_state.messages.append({"role": "assistant", "content": response})
+            resp = get_bot_response(prompt)
+        st.markdown(resp)
+    st.session_state.messages.append({"role":"assistant","content":resp})
 
 st.markdown("---")
-st.markdown('<div style="text-align: center; color: grey;">Developed by <b>Aswinprasath V</b> | Supported by <b>GUVI</b></div>', unsafe_allow_html=True)
+st.markdown('<div style="text-align:center;color:grey;">Developed by <b>Aswinprasath V</b></div>', unsafe_allow_html=True)