Update src/mistral_llm.py

src/mistral_llm.py

@@ -1,28 +1,33 @@
+import streamlit as st
 from transformers import pipeline
 
-# ✅ Light Summarizer — Fast even on CPU
-summarizer = pipeline(
-    "summarization",
-    model="sshleifer/distilbart-cnn-12-6",
-    tokenizer="sshleifer/distilbart-cnn-12-6"
-)
+@st.cache_resource
+def load_summarizer():
+    return pipeline(
+        "summarization",
+        model="sshleifer/distilbart-cnn-12-6",
+        tokenizer="sshleifer/distilbart-cnn-12-6"
+    )
 
-# ✅ Sentiment Analyzer — Fast
-sentiment_pipe = pipeline(
-    "sentiment-analysis",
-    model="distilbert-base-uncased-finetuned-sst-2-english"
-)
+@st.cache_resource
+def load_sentiment():
+    return pipeline(
+        "sentiment-analysis",
+        model="distilbert-base-uncased-finetuned-sst-2-english"
+    )
 
-# 🔍 Fake news detection (rule-based)
-def detect_fake_news(texts):
-    fake_scores = []
-    keywords = ["clickbait", "unverified", "hoax", "rumor", "scam", "misleading"]
-    for text in texts:
-        score = sum(1 for k in keywords if k in text.lower())
-        fake_scores.append(f"{score}/6")
-    return fake_scores
+@st.cache_resource
+def load_fake_news_detector():
+    return pipeline(
+        "text-classification",
+        model="mrm8488/bert-tiny-finetuned-fake-news-detection",
+        tokenizer="mrm8488/bert-tiny-finetuned-fake-news-detection"
+    )
+
+summarizer = load_summarizer()
+sentiment_pipe = load_sentiment()
+fake_news_pipe = load_fake_news_detector()
 
-# ✍️ Summarize list of texts
 def summarize_texts(texts):
     summaries = []
     for text in texts:
@@ -33,7 +38,6 @@ def summarize_texts(texts):
             summaries.append("⚠️ Summary failed")
     return summaries
 
-# 📊 Analyze sentiment
 def analyze_sentiment(texts):
     results = []
     for text in texts:
@@ -43,3 +47,15 @@ def analyze_sentiment(texts):
         except Exception:
             results.append("Unknown")
     return results
+
+def detect_fake_news(texts):
+    results = []
+    for text in texts:
+        try:
+            prediction = fake_news_pipe(text[:512])[0]
+            label = prediction["label"]
+            score = prediction["score"]
+            results.append(f"{label} ({score:.2f})")
+        except Exception:
+            results.append("Unknown")
+    return results