Update app.py

app.py

@@ -1,17 +1,12 @@
 # app.py
 import torch
-from transformers import AutoTokenizer, AutoModelForCausalLM, BitsAndBytesConfig
+from transformers import AutoTokenizer, AutoModelForCausalLM, BitsAndBytesConfig, pipeline  # ← pipeline is here!
 from peft import PeftModel
 import gradio as gr
-import os
-
-# THIS IS THE ONLY NEW LINE YOU NEED
-from huggingface_hub import login
-login(token=os.environ["HF_TOKEN"])   # ← This authenticates the Space
 
 # ===================================
-BASE_MODEL = "meta-llama/Meta-Llama-3-8B-Instruct"
-LORA_ADAPTER = "rishu834763/java-explainer-lora"   # your LoRA
+BASE_MODEL = "meta-llama/Meta-Llama-3.1-8B-Instruct"  # Open, no gate!
+LORA_ADAPTER = "rishu834763/java-explainer-lora"
 
 quantization_config = BitsAndBytesConfig(
     load_in_4bit=True,
@@ -20,7 +15,7 @@ quantization_config = BitsAndBytesConfig(
     bnb_4bit_use_double_quant=True,
 )
 
-print("Loading base model (Llama-3-8B-Instruct 4-bit)...")
+print("Loading Llama-3.1-8B-Instruct 4-bit + your LoRA...")
 base_model = AutoModelForCausalLM.from_pretrained(
     BASE_MODEL,
     quantization_config=quantization_config,
@@ -29,16 +24,12 @@ base_model = AutoModelForCausalLM.from_pretrained(
     trust_remote_code=True,
 )
 
-print("Loading your LoRA adapter...")
 model = PeftModel.from_pretrained(base_model, LORA_ADAPTER)
-
 tokenizer = AutoTokenizer.from_pretrained(BASE_MODEL, trust_remote_code=True)
 tokenizer.pad_token = tokenizer.eos_token
 
-# ===================================
-# Rest of the code stays exactly the same
-# ===================================
-pipe = torch.pipeline(
+# ← FIXED: pipeline from transformers, not torch
+pipe = pipeline(
     "text-generation",
     model=model,
     tokenizer=tokenizer,
@@ -50,7 +41,7 @@ pipe = torch.pipeline(
     return_full_text=False,
 )
 
-SYSTEM_PROMPT = "You are an expert Java teacher. Explain concepts clearly, provide code examples, and answer concisely but completely."
+SYSTEM_PROMPT = "You are an expert Java teacher. Explain concepts clearly with code examples."
 
 def chat(message: str, history):
     messages = [{"role": "system", "content": SYSTEM_PROMPT}]
@@ -61,17 +52,17 @@ def chat(message: str, history):
     messages.append({"role": "user", "content": message})
 
     prompt = tokenizer.apply_chat_template(messages, tokenize=False, add_generation_prompt=True)
-    output = pipe(prompt)[0]["generated_text"]
-    return output
+    outputs = pipe(prompt)
+    return outputs[0]["generated_text"]
 
 # ===================================
-with gr.Blocks(theme=gr.themes.Soft(), title="Java Explainer") as demo:
-    gr.Markdown("# Java Explainer\nPowered by **rishu834763/java-explainer-lora** + Llama-3-8B")
+with gr.Blocks(theme=gr.themes.Soft()) as demo:
+    gr.Markdown("# Java Explainer\nPowered by your LoRA on Llama-3.1-8B-Instruct (4-bit)")
     chatbot = gr.Chatbot(height=620)
-    msg = gr.Textbox(placeholder="Ask anything about Java...", label="Question", container=False)
+    msg = gr.Textbox(placeholder="Ask anything about Java...", container=False)
 
     with gr.Row():
-        send = gr.Button("Send", variant="primary")
+        send = gr.Button("Send 🚀", variant="primary")
         clear = gr.Button("Clear")
 
     send.click(chat, [msg, chatbot], [msg, chatbot]).then(lambda: "", outputs=msg)