| |
| import streamlit as st |
| import torch |
| import transformers |
|
|
| |
| try: |
| from transformers import pipeline |
| except Exception: |
| from transformers.pipelines import pipeline |
|
|
| st.set_page_config(page_title="GUVI AI Assistant", page_icon="🤖", layout="wide") |
|
|
| @st.cache_resource |
| def load_models(): |
| try: |
| 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"Model 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", "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" |
| } |
|
|
| 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 msg in st.session_state.messages: |
| with st.chat_message(msg["role"]): |
| st.markdown(msg["content"]) |
|
|
| SYSTEM_PROMPT = ( |
| "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") |
|
|
| 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 = [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","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() |
|
|
| def get_bot_response(user_input): |
| if len(user_input.split()) < 3: |
| return "Please ask a more specific question." |
| try: |
| 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 = MODEL_LANG_MAP.get(lang2, "en_XX") |
| if src == "en_XX": |
| en_q = user_input |
| else: |
| try: |
| en_q = translator(user_input, src_lang=src, tgt_lang="en_XX")[0].get("translation_text", user_input) |
| except Exception: |
| en_q = user_input |
| prompt = build_prompt(en_q) |
| try: |
| 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: |
| en_resp = str(out[0]) if isinstance(out, list) else str(out) |
| except Exception as e: |
| st.error(f"Generation error: {e}") |
| 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) |
| with st.chat_message("assistant"): |
| with st.spinner("Thinking..."): |
| 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></div>', unsafe_allow_html=True) |
|
|
