Update app.py

app.py

@@ -9,16 +9,18 @@ You can also use efog 🌬️🌁🌫️SqlCoder by cloning this space. 🧬🔬
 Join us : 🌟TeamTonic🌟 is always making cool demos! Join our active builder's🛠️community 👻[![Let's build the future of AI together! 🚀🤖](https://discordapp.com/api/guilds/1109943800132010065/widget.png)](https://discord.gg/GWpVpekp) On 🤗Huggingface: [TeamTonic](https://huggingface.co/TeamTonic) & [MultiTransformer](https://huggingface.co/MultiTransformer) On 🌐Github: [Polytonic](https://github.com/tonic-ai) & contribute to 🌟 [Poly](https://github.com/tonic-ai/poly) 🤗Big thanks to Yuvi Sharma and all the folks at huggingface for the community grant 🤗
 """
 
+global_tokenizer, global_model = None, None
+
 def load_tokenizer_model(model_name):
-    tokenizer = AutoTokenizer.from_pretrained(model_name)
-    model = AutoModelForCausalLM.from_pretrained(
+    global global_tokenizer, global_model
+    global_tokenizer = AutoTokenizer.from_pretrained(model_name)
+    global_model = AutoModelForCausalLM.from_pretrained(
         model_name,
         trust_remote_code=True,
         torch_dtype=torch.float16,
         device_map="auto",
         use_cache=True,
     )
-    return tokenizer, model
 
 def generate_prompt(question, prompt_file="prompt.md", metadata_file="metadata.sql"):
     with open(prompt_file, "r") as f:
@@ -33,14 +35,15 @@ def generate_prompt(question, prompt_file="prompt.md", metadata_file="metadata.s
     return prompt
 
 @spaces.GPU
-def run_inference(question, model, tokenizer):
-    model.to('cuda')
+def run_inference(question):
+    global global_tokenizer, global_model
+    global_model.to('cuda')
     prompt = generate_prompt(question)
-    eos_token_id = tokenizer.eos_token_id
+    eos_token_id = global_tokenizer.eos_token_id
     pipe = pipeline(
         "text-generation",
-        model=model,
-        tokenizer=tokenizer,
+        model=global_model,
+        tokenizer=global_tokenizer,
         max_new_tokens=300,
         do_sample=False,
         num_beams=5,
@@ -62,14 +65,14 @@ def run_inference(question, model, tokenizer):
 
 def main():
     model_name = "defog/sqlcoder2"
-    tokenizer, model = load_tokenizer_model(model_name)
+    load_tokenizer_model(model_name)
 
     with gr.Blocks() as demo:
         gr.Markdown(title)
         question = gr.Textbox(label="Enter your question")
         submit = gr.Button("Generate SQL Query")
         output = gr.Textbox(label="🌬️🌁🌫️SqlCoder-2")
-        submit.click(fn=lambda x: run_inference(x, model, tokenizer), inputs=question, outputs=output)
+        submit.click(fn=run_inference, inputs=question, outputs=output)
 
     demo.launch()