Create app.py

app.py

@@ -0,0 +1,53 @@
+import torch
+from transformers import AutoModelForCausalLM, AutoTokenizer, BitsAndBytesConfig
+from peft import PeftModel
+import gradio as gr
+
+# --- Load Model & Tokenizer ---
+
+base_model_name = "unsloth/llama-3.2-3b-bnb-4bit"
+adapter_model_name = "aismaanly/ai_synthetic"
+
+bnb_config = BitsAndBytesConfig(
+    load_in_4bit=True,
+    bnb_4bit_quant_type="nf4",
+    bnb_4bit_use_double_quant=True,
+    bnb_4bit_compute_dtype=torch.bfloat16
+)
+
+print("Loading base model...")
+model = AutoModelForCausalLM.from_pretrained(
+    base_model_name,
+    quantization_config=bnb_config,
+    device_map="auto"
+)
+
+print("Loading tokenizer...")
+tokenizer = AutoTokenizer.from_pretrained(base_model_name)
+
+print("Loading PEFT adapter...")
+model = PeftModel.from_pretrained(model, adapter_model_name)
+model = model.merge_and_unload()
+print("Model ready!")
+
+# --- Gradio Function ---
+
+def generate_text(prompt):
+    inputs = tokenizer(prompt, return_tensors="pt").to(model.device)
+    outputs = model.generate(**inputs, max_new_tokens=100)
+    text = tokenizer.decode(outputs[0], skip_special_tokens=True)
+    return text
+
+# --- Gradio Interface ---
+def chat_fn(message, history):
+    prompt = message
+    inputs = tokenizer(prompt, return_tensors="pt").to(model.device)
+    outputs = model.generate(**inputs, max_new_tokens=100)
+    text = tokenizer.decode(outputs[0], skip_special_tokens=True)
+    return text
+
+gr.ChatInterface(
+    fn=chat_fn,
+    title="LLM Finetuned Comment Generator",
+    description="Chat with the model.",
+).launch(share=True)