Update app.py

app.py

@@ -25,10 +25,10 @@ simple_instruction = (
     " (например: доступ, платежи, безопасность и т.д.)."
 )
 
-# Модели
+# Лёгкие и реальные модели, подходящие под требования
 models = {
-    "ChatGPT-like (saiga_mistral)": pipeline("text-generation", model="IlyaGusev/saiga_mistral_7b_merged", tokenizer="IlyaGusev/saiga_mistral_7b_merged", device=-1),
-    "DeepSeek-like (ruGPT3-medium)": pipeline("text-generation", model="ai-forever/rugpt3medium_based_on_gpt2", tokenizer="ai-forever/rugpt3medium_based_on_gpt2", device=-1),
+    "ChatGPT-like (ruGPT3-medium)": pipeline("text-generation", model="cointegrated/rugpt3-medium", tokenizer="cointegrated/rugpt3-medium", device=-1),
+    "DeepSeek-like (ruGPT3-small)": pipeline("text-generation", model="ai-forever/rugpt3small_based_on_gpt2", tokenizer="ai-forever/rugpt3small_based_on_gpt2", device=-1),
     "GigaChat-like (rubert-tiny2)": pipeline("text-generation", model="cointegrated/rubert-tiny2", tokenizer="cointegrated/rubert-tiny2", device=-1),
 }
 
@@ -57,13 +57,13 @@ def generate_dual_answers(user_input):
     for name, pipe in models.items():
         # CoT
         start_cot = time.time()
-        out_cot = pipe(prompt_cot, max_new_tokens=300, do_sample=True, top_p=0.9, temperature=0.7)[0]["generated_text"]
+        out_cot = pipe(prompt_cot, max_new_tokens=200, do_sample=True, top_p=0.9, temperature=0.7)[0]["generated_text"]
         end_cot = round(time.time() - start_cot, 2)
         answer_cot = out_cot.strip().split('\n')[-1]
 
         # Simple
         start_simple = time.time()
-        out_simple = pipe(prompt_simple, max_new_tokens=300, do_sample=True, top_p=0.9, temperature=0.7)[0]["generated_text"]
+        out_simple = pipe(prompt_simple, max_new_tokens=150, do_sample=True, top_p=0.9, temperature=0.7)[0]["generated_text"]
         end_simple = round(time.time() - start_simple, 2)
         answer_simple = out_simple.strip().split('\n')[-1]
 
@@ -74,34 +74,29 @@ def generate_dual_answers(user_input):
             "simple_time": end_simple
         }
 
-    return tuple(
-        [
-            results[m]["cot_answer"], f"{results[m]['cot_time']} сек",
-            results[m]["simple_answer"], f"{results[m]['simple_time']} сек"
-        ]
-        for m in models
-    )[0] + tuple(
-        [
-            results[m]["cot_answer"], f"{results[m]['cot_time']} сек",
-            results[m]["simple_answer"], f"{results[m]['simple_time']} сек"
-        ]
-        for m in list(models)[1:]
+    return (
+        results["ChatGPT-like (ruGPT3-medium)"]["cot_answer"], f"{results['ChatGPT-like (ruGPT3-medium)']['cot_time']} сек",
+        results["ChatGPT-like (ruGPT3-medium)"]["simple_answer"], f"{results['ChatGPT-like (ruGPT3-medium)']['simple_time']} сек",
+        results["DeepSeek-like (ruGPT3-small)"]["cot_answer"], f"{results['DeepSeek-like (ruGPT3-small)']['cot_time']} сек",
+        results["DeepSeek-like (ruGPT3-small)"]["simple_answer"], f"{results['DeepSeek-like (ruGPT3-small)']['simple_time']} сек",
+        results["GigaChat-like (rubert-tiny2)"]["cot_answer"], f"{results['GigaChat-like (rubert-tiny2)']['cot_time']} сек",
+        results["GigaChat-like (rubert-tiny2)"]["simple_answer"], f"{results['GigaChat-like (rubert-tiny2)']['simple_time']} сек"
     )
 
 # Интерфейс Gradio
 with gr.Blocks() as demo:
-    gr.Markdown("## 🏦 Сравнение моделей (Классификация клиентских обращений): ChatGPT-like, DeepSeek-like, GigaChat-like")
+    gr.Markdown("## 🏦 Сравнение моделей: Классификация клиентских обращений (CoT и обычный)")
 
-    inp = gr.Textbox(label="Вопрос клиента", placeholder="Например: Я не могу попасть в личный кабинет", lines=2)
-    btn = gr.Button("Сгенерировать")
+    inp = gr.Textbox(label="Обращение клиента", placeholder="Например: Не могу войти в приложение", lines=2)
+    btn = gr.Button("Получить ответы")
 
-    gr.Markdown("### ChatGPT-like (saiga_mistral)")
+    gr.Markdown("### ChatGPT-like (ruGPT3-medium)")
     cot1 = gr.Textbox(label="CoT ответ")
     cot1_time = gr.Textbox(label="Время CoT")
     simple1 = gr.Textbox(label="Обычный ответ")
     simple1_time = gr.Textbox(label="Время обычного")
 
-    gr.Markdown("### DeepSeek-like (ruGPT3-medium)")
+    gr.Markdown("### DeepSeek-like (ruGPT3-small)")
     cot2 = gr.Textbox(label="CoT ответ")
     cot2_time = gr.Textbox(label="Время CoT")
     simple2 = gr.Textbox(label="Обычный ответ")