Update app.py

app.py

@@ -18,10 +18,11 @@ def load_pipelines():
         "image-to-text",
         model="Salesforce/blip-image-captioning-base"
     )
-    # 2) Story generation with Flan-T5
+    # 2) Story generation with a bigger Flan-T5
     storyteller = pipeline(
         "text2text-generation",
-        model="google/flan-t5-base"
+        model="google/flan-t5-large",
+        device=0  # set to -1 if you only have CPU
     )
     return captioner, storyteller
 
@@ -36,22 +37,21 @@ if uploaded:
     # —––––––– 1. Caption
     with st.spinner("🔍 Looking at the image..."):
         cap_outputs = captioner(image)
-        # BLIP returns a list of dicts with key "generated_text"
         cap = cap_outputs[0].get("generated_text", "").strip()
     st.markdown(f"**Caption:** {cap}")
 
     # —––––––– 2. Story generation
     prompt = (
-        "Write a playful, 50–100 word story for 3–10 year-old children "
+        "Write a playful, imaginative story of about 50–100 words for 3–10 year-olds, "
         f"based on this description:\n\n“{cap}”\n\nStory:"
     )
     with st.spinner("✍️ Writing a story..."):
         out = storyteller(
             prompt,
-            max_length=200,
+            max_length=250,       # give it a bit more room
             do_sample=True,
-            top_p=0.9,
-            temperature=0.8,
+            top_p=0.95,
+            temperature=0.7,
             num_return_sequences=1
         )
         story = out[0]["generated_text"].strip()
@@ -65,3 +65,4 @@ if uploaded:
         tts.write_to_fp(tmp)
         tmp.flush()
     st.audio(tmp.name, format="audio/mp3")
+