Update app.py

app.py

@@ -1,4 +1,4 @@
-# app.py – CPU-only, ultra-stable, works on any HF Space (free tier included)
+# app.py – FINAL VERSION (November 2025) – Instant output, dual input
 import torch
 from transformers import AutoTokenizer, AutoModelForCausalLM, pipeline
 from peft import PeftModel
@@ -7,61 +7,100 @@ import gradio as gr
 BASE_MODEL   = "mistralai/Mistral-7B-Instruct-v0.2"
 LORA_ADAPTER = "rishu834763/java-explainer-lora"
 
-print("Loading Mistral-7B-Instruct-v0.2 + your Java LoRA (8-bit CPU mode)...")
+print("Loading your Java Explainer (8-bit CPU mode – super fast & stable)...")
 
-# 8-bit on CPU = ~7–8 GB RAM, fits perfectly on every HF Space (even free CPU)
+# 8-bit CPU = perfect balance: fast, low RAM, no CUDA needed
 model = AutoModelForCausalLM.from_pretrained(
     BASE_MODEL,
-    device_map="auto",           # will automatically use CPU if no GPU
-    load_in_8bit=True,           # ← 8-bit instead of 4-bit → no CUDA driver needed
+    load_in_8bit=True,
+    device_map="auto",
     torch_dtype=torch.float16,
     low_cpu_mem_usage=True,
 )
 
-# Apply your LoRA (adds only ~168 MB)
 model = PeftModel.from_pretrained(model, LORA_ADAPTER)
-
 tokenizer = AutoTokenizer.from_pretrained(BASE_MODEL)
 tokenizer.pad_token = tokenizer.eos_token
 
+# Fast pipeline settings for instant response
 pipe = pipeline(
     "text-generation",
     model=model,
     tokenizer=tokenizer,
     max_new_tokens=1024,
-    temperature=0.3,
+    temperature=0.2,
     top_p=0.95,
     do_sample=True,
-    repetition_penalty=1.15,
+    repetition_penalty=1.18,
     return_full_text=False,
+    eos_token_id=tokenizer.eos_token_id,
+    pad_token_id=tokenizer.pad_token_id,
 )
 
-SYSTEM_PROMPT = "You are an expert Java teacher. Always explain clearly with clean, runnable code examples."
-
-def chat(message: str, history):
-    messages = [{"role": "system", "content": SYSTEM_PROMPT}]
-    for user, assistant in history:
-        messages.append({"role": "user", "content": user})
-        if assistant:
-            messages.append({"role": "assistant", "content": assistant})
-    messages.append({"role": "user", "content": message})
+SYSTEM_PROMPT = """You are the world's best Java teacher. 
+Always respond with:
+• Clear explanation
+• Clean, runnable, modern Java code (Java 17+)
+• Best practices (records, var, sealed classes, etc.)
+• Fix any bugs or bad patterns
+Never say "I can't see the code" — always assume it's provided."""
 
+def generate(instruction: str, code: str = ""):
+    user_input = f"### Instruction:\n{instruction.strip()}\n\n### Code (if any):\n{code.strip()}" if code.strip() else instruction.strip()
+    
+    messages = [
+        {"role": "system", "content": SYSTEM_PROMPT},
+        {"role": "user",   "content": user_input}
+    ]
+    
     prompt = tokenizer.apply_chat_template(messages, tokenize=False, add_generation_prompt=True)
-    output = pipe(prompt)[0]["generated_text"]
+    
+    output = pipe(prompt, max_new_tokens=1024)[0]["generated_text"].strip()
     return output
 
-# UI
-with gr.Blocks(theme=gr.themes.Soft(), title="Java Explainer") as demo:
-    gr.Markdown("# Java Explainer Pro\nPowered by **rishu834763/java-explainer-lora** + Mistral-7B-v0.2 (CPU 8-bit)")
-    chatbot = gr.Chatbot(height=620)
-    msg = gr.Textbox(placeholder="Ask anything about Java...", container=False)
+# Beautiful dual-input UI
+with gr.Blocks(theme=gr.themes.Soft(), title="Java Explainer Pro") as demo:
+    gr.Markdown("# Java Explainer Pro\nAsk anything — explain, fix, improve, teach")
 
     with gr.Row():
-        send   = gr.Button("Send", variant="primary")
-        clear  = gr.Button("Clear")
+        with gr.Column(scale=1):
+            instruction = gr.Textbox(
+                label="Instruction / Question",
+                placeholder="e.g. Explain this code / Fix this bug / Convert to Java records / Make it thread-safe / Best way to read a file in Java 17",
+                lines=6
+            )
+            code_input = gr.Code(
+                label="Java Code (optional)",
+                language="java",
+                lines=12,
+                placeholder="// Paste your Java code here (or leave empty)"
+            )
+            
+            with gr.Row():
+                submit = gr.Button("Explain / Fix / Improve", variant="primary", size="lg")
+                clear = gr.Button("Clear")
+
+        with gr.Column(scale=1):
+            output = gr.Markdown(label="Answer")
+
+    # Instant generation
+    submit.click(
+        fn=generate,
+        inputs=[instruction, code_input],
+        outputs=output
+    )
+    
+    # Also allow Enter key
+    instruction.submit(
+        fn=generate,
+        inputs=[instruction, code_input],
+        outputs=output
+    )
 
-    send.click(chat, [msg, chatbot], [msg, chatbot]).then(lambda: "", outputs=msg)
-    msg.submit(chat, [msg, chatbot], [msg, chatbot]).then(lambda: "", outputs=msg)
-    clear.click(lambda: None, None, chatbot, queue=False)
+    clear.click(lambda: ("", "", ""), None, [instruction, code_input, output])
 
-demo.queue(max_size=100).launch(server_name="0.0.0.0", server_port=7860)
+demo.queue(max_size=20).launch(
+    server_name="0.0.0.0",
+    server_port=7860,
+    share=True
+)