Update app110.py

app110.py

@@ -378,15 +378,43 @@ if "task_choice" in st.session_state:
         
         user_prompt = st.text_area("User Prompt (optional)")
 
+        # # Updated prompt template including system role
+        # prompt_template = PromptTemplate(
+        #     input_variables=["system_role", "classification_type", "domain", "num_examples", 
+        #                    "min_words", "max_words", "labels", "user_prompt", "few_shot_examples", "additional_attributes"],
+        #     template=(
+        #        "{system_role}\n"
+        #          "- Use the following parameters:\n"
+        #         "- Generate {num_examples} examples\n"
+        #         "- Each example should be between {min_words} to {max_words} words long\n"
+        #         "- Use these labels: {labels}.\n"
+        #         "- Use the following additional attributes:\n"
+        #         "- {additional_attributes}\n"
+        #         "- Generate the examples in this format: 'Example text. Label: label'\n"
+        #         "- Do not include word counts or any additional information\n"
+        #         "- Always use your creativity and intelligence to generate unique and diverse text data\n"
+        #         "- In sentiment analysis, ensure that the sentiment classification is clearly identified as Positive, Negative, or Neutral. Do not leave the sentiment ambiguous.\n"
+        #         "- In binary sentiment analysis, classify text strictly as either Positive or Negative. Do not include or imply Neutral as an option.\n"
+        #         "- Write unique examples every time.\n"
+        #         "- DO NOT REPEAT your gnerated text. \n"
+        #         "- For each Output, describe it once and move to the next.\n"
+        #         "- List each Output only once, and avoid repeating details.\n"
+        #         "- Additional instructions: {user_prompt}\n\n"
+        #         "- Use the following examples as a reference in the generation process\n\n {few_shot_examples}. \n"
+        #         "- Think step by step, generate numbered examples, and check each newly generated example to ensure it has not been generated before. If it has, modify it"
+                
+        #      )
+        #  )
+
         # Updated prompt template including system role
         prompt_template = PromptTemplate(
             input_variables=["system_role", "classification_type", "domain", "num_examples", 
                            "min_words", "max_words", "labels", "user_prompt", "few_shot_examples", "additional_attributes"],
-            template=(
+        template=(
                "{system_role}\n"
-                 "- Use the following parameters:\n"
+                "- Use the following parameters:\n"
                 "- Generate {num_examples} examples\n"
-                "- Each example should be between {min_words} to {max_words} words long\n"
+                "- Each example should be between {min_words} to {max_words} words long, 50 words short text and more than 50 words long text with not less than 4 lines\n"
                 "- Use these labels: {labels}.\n"
                 "- Use the following additional attributes:\n"
                 "- {additional_attributes}\n"
@@ -396,9 +424,7 @@ if "task_choice" in st.session_state:
                 "- In sentiment analysis, ensure that the sentiment classification is clearly identified as Positive, Negative, or Neutral. Do not leave the sentiment ambiguous.\n"
                 "- In binary sentiment analysis, classify text strictly as either Positive or Negative. Do not include or imply Neutral as an option.\n"
                 "- Write unique examples every time.\n"
-                "- DO NOT REPEAT your gnerated text. \n"
-                "- For each Output, describe it once and move to the next.\n"
-                "- List each Output only once, and avoid repeating details.\n"
+                "- DO NOT REPEAT gnerated examples. \n"
                 "- Additional instructions: {user_prompt}\n\n"
                 "- Use the following examples as a reference in the generation process\n\n {few_shot_examples}. \n"
                 "- Think step by step, generate numbered examples, and check each newly generated example to ensure it has not been generated before. If it has, modify it"