Update app.py

Add spaces.GPU decorator

app.py

@@ -2,6 +2,7 @@ from time import sleep
 
 import gradio as gr
 from transformers import AutoModelForCausalLM, AutoTokenizer
+import spaces
 import torch
 from duckduckgo_search import DDGS
 import re
@@ -322,6 +323,11 @@ class Applicant:
     def __init__(self, resume):
         self.resume = resume
 
+@spaces.GPU
+def write(inputs):
+    _output = model.generate(**inputs, max_new_tokens=max_new_tokens, do_sample=True, temperature=0.6, top_k=40, top_p=0.9, repetition_penalty=1.1)
+    return _output
+
 def smol_lm_jd_process(job_description, system_prompt, max_new_tokens=512):
     prompt = f"""<|im_start|>system
 {system_prompt}<|im_end|>
@@ -329,8 +335,8 @@ def smol_lm_jd_process(job_description, system_prompt, max_new_tokens=512):
 {job_description}<|im_end|>
 <|im_start|>assistant
 """
-    inputs = tokenizer(prompt, return_tensors="pt").to(device)
-    output = model.generate(**inputs, max_new_tokens=max_new_tokens, do_sample=True, temperature=0.6, top_k=40, top_p=0.9, repetition_penalty=1.1)
+    inputs = tokenizer(prompt, return_tensors="pt").to(device)   
+    output = write(inputs)
     response = tokenizer.decode(output[0], skip_special_tokens=False)
     start_idx = response.find("<|im_start|>assistant")
     end_idx = response.find("<|im_end|>", start_idx)