using model directly

app.py

@@ -1,12 +1,18 @@
 import torch
-from transformers import pipeline
+from transformers import AutoTokenizer, AutoModelForCausalLM
 import gradio as gr
 
-pipe = pipeline("text-generation", model="MTSAIR/multi_verse_model")
+# Load model directly
+tokenizer = AutoTokenizer.from_pretrained("MTSAIR/multi_verse_model")
+model = AutoModelForCausalLM.from_pretrained("MTSAIR/multi_verse_model")
 
 def greet(name):
-    res = pipe(name, do_sample=True, temperature=0.7, top_k=50, top_p=0.95)
-    return res[0]['generated_text']
+    #i want to get same result res = pipe(name, do_sample=True, temperature=0.7, top_k=50, top_p=0.95)
+    #but using tokenizer and model
+    input_ids = tokenizer.encode(name, return_tensors='pt')
+    res = model.generate(input_ids, do_sample=True, temperature=0.7, top_k=50, top_p=0.95)
+    generated = tokenizer.decode(res[0], skip_special_tokens=True)
+    return generated
 
 iface = gr.Interface(fn=greet, inputs="text", outputs="text")
 iface.launch()