Update app.py

app.py

@@ -1,31 +1,27 @@
 
-from transformers import AutoTokenizer, AutoModelForCausalLM
-import torch
+from transformers import AutoTokenizer, AutoModelForSeq2SeqLM
 import gradio as gr
 
-model_name = "sberbank-ai/rugpt3medium_based_on_gpt2"
+model_name = "cointegrated/rut5-small"
 tokenizer = AutoTokenizer.from_pretrained(model_name)
-model = AutoModelForCausalLM.from_pretrained(model_name)
+model = AutoModelForSeq2SeqLM.from_pretrained(model_name)
 
 def generate_response(prompt):
-    instruction = f"Ответь кратко и ясно на вопрос:\n{prompt.strip()}\nОтвет:"
-    input_ids = tokenizer.encode(instruction, return_tensors="pt")
+    instruction = f"Ответь кратко: {prompt.strip()}"
+    input_ids = tokenizer.encode(instruction, return_tensors="pt", max_length=512, truncation=True)
 
     output = model.generate(
         input_ids,
-        max_new_tokens=100,
+        max_new_tokens=80,
         do_sample=True,
-        top_k=50,
-        top_p=0.95,
-        temperature=0.9,
-        pad_token_id=tokenizer.eos_token_id,
+        top_p=0.9,
+        temperature=0.7,
+        repetition_penalty=1.2,
         eos_token_id=tokenizer.eos_token_id
     )
 
     response = tokenizer.decode(output[0], skip_special_tokens=True)
-    
-    # Удалим часть промпта из начала
-    return response.replace(instruction, "").strip()
+    return response.strip()
 
 iface = gr.Interface(
     fn=generate_response,