commit

app.py

@@ -1,156 +1,60 @@
 import gradio as gr
-import os
-import re
-from transformers import pipeline
-from openai import OpenAI
+import openai
 
-# Initialize the title and description
-title = "System Prompt Depersonaliser"
+title = "System Prompt Depersonalizer"
 description = """
-This app converts personalized LLM system prompts into depersonalized versions by removing personal elements 
-while preserving the core functionality. Upload your personalized system prompt, and get a generalized version 
-that can be shared with and used by a wider audience.
+This app transforms personalized system prompts into generalized versions. Enter a prompt and an OpenAI API key.
 """
 
-# Function to depersonalize using Hugging Face model
-def depersonalize_with_hf(personalized_prompt):
+def depersonalize_with_openai(prompt, api_key):
     try:
-        # Using a text generation pipeline with a suitable model
-        generator = pipeline('text2text-generation', model='google/flan-t5-base', max_length=1024)
-        
-        # Create a prompt that instructs the model what to do
-        instruction = """
-        Transform the following personalized system prompt into a generalized version by:
-        1. Removing personal elements (names, specific hardware/software, locations)
-        2. Replacing personal references with generic alternatives
-        3. Broadening specific technical requirements when appropriate
-        4. Maintaining the core functionality and purpose
-        5. Preserving the overall structure and flow
-
-        Personalized prompt:
-        """
-        
-        full_prompt = instruction + personalized_prompt
-        
-        # Generate the depersonalized version
-        result = generator(full_prompt, max_length=1024)
-        depersonalized = result[0]['generated_text']
-        
-        return depersonalized
-    except Exception as e:
-        return f"Error using Hugging Face model: {str(e)}\nPlease try using OpenAI instead."
-
-# Function to depersonalize using OpenAI
-def depersonalize_with_openai(personalized_prompt, api_key):
-    try:
-        client = OpenAI(api_key=api_key)
-        
-        response = client.chat.completions.create(
+        openai.api_key = api_key
+        response = openai.ChatCompletion.create(
             model="gpt-3.5-turbo",
             messages=[
                 {"role": "system", "content": """
-                You are an AI assistant specializing in transforming personalized system prompts into generalized versions
-                that can be shared with and used by a wider audience. Your task is to identify and remove personal elements
-                while preserving the core functionality and purpose of the original prompt.
-
-                Task Breakdown:
-                1. Analyze the Original Prompt
-                   - Identify personalized elements such as:
-                     * Names (e.g., Daniel Rosehill)
-                     * Specific hardware or software configurations
-                     * Location-specific references
-                     * Unique use cases or workflows
-                     * Personal preferences or requirements
-
-                2. Generalize the Content
-                   - Replace personal references with generic alternatives
-                   - Broaden specific technical requirements when appropriate
-                   - Maintain the core functionality and purpose
-                   - Preserve the overall structure and flow of instructions
-
-                3. Maintain Quality
-                   While generalizing the prompt:
-                   - Preserve clear instructions and constraints
-                   - Keep specialized knowledge and capabilities
-                   - Ensure the prompt remains coherent and effective
-                   - Retain unique value propositions of the original
-
-                Provide only the depersonalized system prompt without any additional explanations or formatting.
-                """},
-                {"role": "user", "content": personalized_prompt}
+You are an AI assistant specializing in transforming personalized system prompts into generalized versions that can be shared with and used by a wider audience. Your task is to identify and remove personal elements while preserving the core functionality and purpose of the original prompt.
+
+Task Breakdown:
+1. Analyze the Original Prompt
+   - Identify personalized elements such as:
+     * Names (e.g., Daniel Rosehill)
+     * Specific hardware or software configurations
+     * Location-specific references
+     * Unique use cases or workflows
+     * Personal preferences or requirements
+
+2. Generalize the Content
+   - Replace personal references with generic alternatives
+   - Broaden specific technical requirements when appropriate
+   - Maintain the core functionality and purpose
+   - Preserve the overall structure and flow of instructions
+
+Provide only the depersonalized system prompt without any additional explanations or formatting.
+"""}, 
+                {"role": "user", "content": prompt}
             ],
-            max_tokens=1500
+            max_tokens=1200
         )
-        
-        return response.choices[0].message.content
+        return response.choices[0].message["content"]
     except Exception as e:
-        return f"Error using OpenAI: {str(e)}"
+        return f"Error: {str(e)}"
 
-# Main function to process the input
-def process_prompt(personalized_prompt, use_openai, api_key=""):
-    if not personalized_prompt:
-        return "Please enter a personalized system prompt."
-    
-    if use_openai and not api_key:
-        return "Please provide an OpenAI API key to use OpenAI for processing."
-    
-    if use_openai:
-        return depersonalize_with_openai(personalized_prompt, api_key)
-    else:
-        return depersonalize_with_hf(personalized_prompt)
-
-# Create the Gradio interface
 with gr.Blocks() as demo:
     gr.Markdown(f"# {title}")
     gr.Markdown(description)
-    
-    with gr.Row():
-        with gr.Column():
-            input_text = gr.Textbox(
-                label="Personalized System Prompt",
-                placeholder="Enter your personalized system prompt here...",
-                lines=10
-            )
-        
-        with gr.Column():
-            output_text = gr.Textbox(
-                label="Depersonalized System Prompt",
-                lines=10
-            )
-    
-    with gr.Row():
-        use_openai = gr.Checkbox(label="Use OpenAI (recommended for better results)")
-    
-    with gr.Row():
-        api_key_input = gr.Textbox(
-            label="OpenAI API Key (only needed if 'Use OpenAI' is checked)",
-            placeholder="sk-...",
-            type="password",
-            visible=True
-        )
-    
-    with gr.Row():
-        submit_btn = gr.Button("Depersonalize")
-    
-    # Set up the click event
-    submit_btn.click(
-        fn=process_prompt,
-        inputs=[input_text, use_openai, api_key_input],
-        outputs=output_text
-    )
-    
-    # Add example inputs
-    gr.Examples(
-        [
-            [
-                "Your task is to help Daniel identify the best way to backup his Open SUSE Linux desktop on his local network. On the local network Daniel has a Synology NAS. You should prioritise direct guidance for incremental backup tools."
-            ],
-            [
-                "You are an AI assistant for Sarah Johnson, a graphic designer who uses Adobe Creative Suite on her MacBook Pro. Help Sarah organize her design files and provide tips for optimizing her workflow in Photoshop and Illustrator. Sarah prefers dark mode interfaces and needs to prepare files for both print and web."
-            ]
-        ],
-        inputs=[input_text]
+
+    prompt_input = gr.Textbox(label="Personalized System Prompt", lines=10)
+    api_key_input = gr.Textbox(label="OpenAI API Key", type="password")
+
+    output_box = gr.Textbox(label="Depersonalized Prompt", lines=10)
+
+    run_btn = gr.Button("Depersonalize")
+
+    run_btn.click(
+        depersonalize_with_openai,
+        inputs=[prompt_input, api_key_input],
+        outputs=output_box
     )
 
-# Launch the app
-demo.launch()
+demo.launch()

requirements.txt

@@ -1,4 +1,2 @@
-gradio==5.29.0
-transformers
-torch
-openai
+gradio
+openai

sp.md

@@ -1,36 +0,0 @@
-"You are an AI assistant specializing in transforming personalized system prompts into generalized versions that can be shared with and used by a wider audience. Your task is to identify and remove personal elements while preserving the core functionality and purpose of the original prompt.
-
-\## Task Breakdown
-1\. Analyze the Original Prompt
-  - Identify personalized elements such as:
-    \* Names (e.g., Daniel Rosehill)
-    \* Specific hardware or software configurations
-    \* Location-specific references
-    \* Unique use cases or workflows
-    \* Personal preferences or requirements
-
-2\. Generalize the Content
-  - Replace personal references with generic alternatives
-  - Broaden specific technical requirements when appropriate
-  - Maintain the core functionality and purpose
-  - Preserve the overall structure and flow of instructions
-
-3\. Example Transformations
-
-\#### Before Depersonalization:
-\`Your task is to help Daniel identify the best way to backup his Open SUSE Linux desktop on his local network. On the local network Daniel has a Synology NAS. You should prioritise direct guidance for incremental backup tools.\`
-
-\#### After Depersonalization:
-\`Your task is to assist users in deploying optimal backup strategies over local area networks. Focus on providing direct guidance to help users select suitable backup tools.\`
-
-4\. Maintain Quality
-  While generalizing the prompt:
-  - Preserve clear instructions and constraints
-  - Keep specialized knowledge and capabilities
-  - Ensure the prompt remains coherent and effective
-  - Retain unique value propositions of the original
-
-\## Output Format
-Provide the depersonalized system prompt in Markdown format within a code fence for easy copying and implementation by the user.
-\`\`\`markdown
-Your task is to assist users in deploying optimal backup strategies over local area networks. Focus on providing direct guidance to help users select suitable backup tools."