Home » Thought Overload? How To Search ChatGPT’s Chat Historical past | by Tristan Wolff | Could, 2023

Thought Overload? How To Search ChatGPT’s Chat Historical past | by Tristan Wolff | Could, 2023

by Narnia
0 comment

A easy resolution to uncover hidden chat messages, prompts, and undiscovered insights

Image by writer & Midjourney

Without a built-in search perform, it may be barely irritating to seek out that valuable gem of knowledge in your ChatGPT chat historical past. Fortunately, there’s a workaround that not solely means that you can delve into your chat historical past but additionally to export your conversations into versatile codecs like HTML and JSON. Ready to discover? Let’s get began!

To start, navigate to chat.openai.com and click on on the menu within the left nook …

… then, select “Settings” and “Data controls” …

…, click on on “Export knowledge” and make sure …

Within a couple of minutes, a obtain hyperlink to your chat historical past will arrive in your e-mail.

The ZIP file you downloaded incorporates 5 information, however for our goal, we’re solely fascinated with two of them:

chat.html

A complete HTML file containing all of your conversations. Load it with a browser (be affected person, it would take some time) and voilà, you may search it like an everyday webpage:

conversations.json

This file incorporates all of your conversations in a machine-readable JSON format, which permits us to make use of our personal search algorithms to discover our chat historical past.

Say, as an example, that you simply wish to uncover your entire prompts that use the phrase “picture caption”. A number of strains of Python code can do the job, and I’ve ready a primary Colab pocket book to information you thru it: Colab Notebook Link

Here’s a sneak peek into the way it works:

  1. Run the “Load JSON file” code block to make an add button seem.

2. Choose “conversations.json” from the ZIP file OpenAI despatched you and add it. This might take a second.

3. Once the add finishes, run the “Search JSON file” code block.

4. Specify whether or not you wish to filter “consumer” or “assistant” messages (or each) and enter a search string. Click “Filter Messages” to execute the search.

5. The filtered messages shall be saved to your digital drive as “filtered_messages.csv”. You can entry the file by clicking on the folder icon.

You may also like

Leave a Comment