Create app.py

app.py

@@ -0,0 +1,44 @@
+import gradio as gr
+from transformers import AutoModelForCausalLM, AutoTokenizer, BitsAndBytesConfig
+from peft import PeftModel
+import torch
+
+print("Loading model...")
+
+bnb_config = BitsAndBytesConfig(
+    load_in_4bit=True,
+    bnb_4bit_quant_type="nf4",
+    bnb_4bit_compute_dtype=torch.float16,
+    bnb_4bit_use_double_quant=True
+)
+
+base_model_name = "unsloth/Llama-3.2-1B-Instruct"
+tokenizer = AutoTokenizer.from_pretrained(base_model_name)
+base_model = AutoModelForCausalLM.from_pretrained(
+    base_model_name,
+    quantization_config=bnb_config,
+    device_map="auto",
+    trust_remote_code=True
+)
+
+model = PeftModel.from_pretrained(base_model, "AA65327/lora_model")
+
+print("Model loaded!")
+
+def classify_emotion(text):
+    prompt = f"Classify the emotion in this text: {text}\n\nEmotion:"
+    inputs = tokenizer(prompt, return_tensors="pt").to(model.device)
+    outputs = model.generate(**inputs, max_new_tokens=50, temperature=0.3, pad_token_id=tokenizer.eos_token_id)
+    response = tokenizer.decode(outputs[0], skip_special_tokens=True)
+    emotion = response.split("Emotion:")[-1].strip().split()[0]
+    return emotion
+
+demo = gr.Interface(
+    fn=classify_emotion,
+    inputs=gr.Textbox(label="Enter text to classify", placeholder="I am so happy today!"),
+    outputs=gr.Textbox(label="Detected Emotion"),
+    title="Emotion Classifier",
+    description="Classify emotions in text using a fine-tuned Llama model"
+)
+
+demo.launch()