Update app.py

app.py

@@ -10,7 +10,7 @@ import gradio as gr
 
 
 tokenizer = AutoTokenizer.from_pretrained("openai-community/gpt2")
-model = AutoModelForCausalLM.from_pretrained("openai-community/gpt2").to(torch_device)
+model = AutoModelForCausalLM.from_pretrained("openai-community/gpt2")
 
 
 # Define functions
@@ -21,7 +21,7 @@ global chosen_strategy
 def generate(input_text, number_steps, number_beams, number_beam_groups, diversity_penalty, length_penalty, num_return_sequences, temperature, no_repeat_ngram_size, repetition_penalty, early_stopping, beam_temperature, top_p, top_k,penalty_alpha,top_p_box,top_k_box,strategy_selected,model_selected):
 
     chosen_strategy = strategy_selected
-    inputs = tokenizer(input_text, return_tensors="pt").to(torch_device)
+    inputs = tokenizer(input_text, return_tensors="pt")
     
     if chosen_strategy == "Sampling":
         
@@ -47,7 +47,7 @@ def generate(input_text, number_steps, number_beams, number_beam_groups, diversi
         beam_temp_flag = beam_temperature
         early_stop_flag = early_stopping
     
-        inputs = tokenizer(input_text, return_tensors="pt").to(torch_device)    
+        inputs = tokenizer(input_text, return_tensors="pt")   
         outputs = model.generate(
             
             **inputs,
@@ -82,7 +82,7 @@ def generate(input_text, number_steps, number_beams, number_beam_groups, diversi
         if number_beam_groups > number_beams:
             number_beams = number_beam_groups
         
-        inputs = tokenizer(input_text, return_tensors="pt").to(torch_device)    
+        inputs = tokenizer(input_text, return_tensors="pt")    
         outputs = model.generate(
             
             **inputs,
@@ -130,12 +130,12 @@ def load_model(model_selected):
     
     if model_selected == "gpt2":
         tokenizer = AutoTokenizer.from_pretrained("gpt2")
-        model = AutoModelForCausalLM.from_pretrained("gpt2", pad_token_id=tokenizer.eos_token_id).to(torch_device)
+        model = AutoModelForCausalLM.from_pretrained("gpt2", pad_token_id=tokenizer.eos_token_id)
         #print (model_selected + " loaded")
     
     if model_selected == "Gemma 2":
         tokenizer = AutoTokenizer.from_pretrained("google/gemma-2b")
-        model = AutoModelForCausalLM.from_pretrained("google/gemma-2b").to(torch_device)
+        model = AutoModelForCausalLM.from_pretrained("google/gemma-2b")