Update app.py

app.py

@@ -1,3 +1,30 @@
 import gradio as gr
 
-gr.Interface.load("models/nitinbhayana/Llama-2-7b-chat-hf-review-phrases-sentiments-v2").launch()
+#gr.Interface.load("models/nitinbhayana/Llama-2-7b-chat-hf-review-phrases-sentiments-v2").launch()
+
+
+from transformers import pipeline
+
+pipeline = pipeline("text-generation", model="nitinbhayana/Llama-2-7b-chat-hf-keyword-category-brand-v1")
+
+def predict(review):
+    
+    prompt=f"""[INST] <<SYS>>
+You are a helpful assistant that provides accurate and concise responses. Do not hallucinate.
+<</SYS>>
+Below is an instruction that describes a task, paired with an input that provides further context. Write a response that appropriately completes the request.
+### Instruction:
+Analyze the following printer product review on amazon to extract any relevant phrases from the review that are related to each of the specified topics: ['connectivity', 'customer_support', 'delivery_and_packaging', 'durability', 'features', 'ink_toner_cost_efficiency', 'noise', 'print_quality', 'setup_and_installation', 'speed', 'user_interface', 'value_for_money', 'portability', 'easy_to_use', 'mobile_printing_functionality','authenticity','product_quality'], and indicate the sentiment expressed in each phrase.The sentiment evaluations should range from 0 to 5, with 5 represents extremely positive sentiment, 4 indicates very positive sentiment, 3 suggests neutral sentiment, 2 reflects somewhat negative sentiment, 1 represents highly negative sentiment, and 0 means not mentioned. Provide the output in JSON format.
+### Input:
+{review}
+[/INST]
+"""
+    predictions = pipeline(prompt)
+    return (predictions)
+
+gr.Interface(
+    predict,
+    inputs='text',
+    outputs='text',
+    title="Printer review categorization with sentiments",
+).launch()