[DAO:9f83a53] ChatGPT DCL-Docs Improvement

[dao]

by 0x1b35f4d22b2c8bc706bcfcc383a1ca77fb4fc600 (punkpink)

Should the following $3,200 grant in the Platform category be approved?

Abstract

ChatGPT DCL-Docs will be improved with all the official documentation of Decentraland.

Grant size

3,200 USD in DAI

Project duration

1 month

Beneficiary address

0x8acb9cD89273fB013c544e51d16Fc383e7Ee216e

Email address

punkpinkdcl@gmail.com

Description

Poll prior to this proposal: Should ChatGPT DCL-Docs be improved?

DCL-Docs is a ChatGPT developed to provide a powerful consultation tool for Decentraland’s documentation, however, the official websites of Decentraland contain much more data that is essential for understanding the platform’s operation from beginning to end. This proposal aims to complete ChatGPT DCL-Docs with all available information from Decentraland as follows.

ChatGPT DCL-Docs will be enhanced with all the official documentation related to the operation and use of Decentraland. This will involve a process of selecting, extracting, processing, and structuring the documentation.
During this procedure, all relevant information for understanding the platform’s use will be considered, avoiding data such as calendars, events, wearables, etc., due to their typically ephemeral interest and daily changes, which are not significantly relevant to the knowledge of the platform’s operation and could negatively interfere with some model responses.

The following websites will be used as entry points to access all published information and their internal links, in other words, relevant information from all sections and subsections of these URLs will be extracted:

Welcome to Decentraland

Decentraland

decentraland-scenes

Decentraland DAO | Instagram | Linktree

This will obviously include sections like Docs, Forum, Blog, and the rest of the sections and subsections of these websites.

External Image

Additional Considerations

• Currently, only ChatGPT Plus users have access to special functions such as Advanced Data Analysis and Browsing. This limits the current chat’s reach to these users.
• Link to ChatGPT DCL-Docs: https://chat.openai.com/g/g-O6MHZQtbo-dcl-docs

Roadmap and milestones

1 - Documentation Selection:

• Identify and select relevant official documentation on Decentraland’s websites (decentraland.org, GitHub, linktr.ee/decentralanddao).
• Focus on information about the operation and use of Decentraland.

2. Data Extraction:

• Gather data from all relevant sections and subsections of the mentioned sites.
• Avoid ephemeral information such as calendars, events, wearables and others.

3. Processing:

• Organize and structure the information collected to make it easier for ChatGPT to access and understand.

4. Integration into ChatGPT DCL-Docs:

• Incorporate the structured documentation into the ChatGPT DCL-Docs system.
• Conduct tests to ensure functionality and accuracy of responses.

5. Launch:

• Launch the updated version of ChatGPT DCL-Docs.

Vote on this proposal on the Decentraland DAO

View this proposal on Snapshot

[PunkPink]

This proposal has been published in the Platform category because DCL-Docs is a tool and at no point will additional educational content be created. All current existing documentation that is processed is in order to instruct ChatGPT to build this application.

[InJesterr]

Easy yes for me, was wondering does it help noobies with coding too?

[PunkPink]

Hello @SugarClub, thank you for your questions, I will clarify them below:

1. Your chatbot has been programmed based on the Browsing function, as you have said this can be done easily without needing to extract or process the information, however with this methodology ChatGPT navigates through multiple websites sometimes without finding the corresponding data:

Example of your chat:

The answer to this question is found here: Report a bug | Decentraland Documentation, however your chat has not been able to locate it because it is parsing URLs instead of the information already processed and structured.

2. Only one, but this one must give reliable answers, which is what this proposal is about.

3. I think I have answered this in answer 1.

4. I think this is also answered in answer 1.

If you have more questions please let me know.

Cheers!

[PunkPink]

Hello @InJesterr, thanks for your question.

Yes, it will be a very useful tool for this.

[InJesterr]

Alright I will definetely tell my community about this, have alot of people with land that want to build but dont want to pay for coding!

[PunkPink]

Of course, allow me to explain.

Adding the Canny URL does not solve this problem, in fact if you read my previous comment carefully I mention that the information that I have requested from your chat was already in docs.decentraland.org and your chat has not been able to locate it.

In fact, while your chat was trying to find the answer to this (despite this information being available in a URL that you have entered), your chat has browsed many other pages that were not even official Decentraland, potentially giving a explanation and not at all reliable. Coincidentally your chat said that no such information was available just as it could have said anything else.

Conclusion: The difference between your chat and the one proposed to be developed with this proposal lies mainly in the structuring of the information, that means that processed information will be introduced to DCL-Docs, not just raw URLs.

I don’t know how I can be more precise in my answer, but feel free to ask again if I haven’t expressed myself well.

Cheers!

[InJesterr]

@SugarClub

here is a boiler plate starter code for making a token gated scene in SDK 7 , you can literally code with it already , provided you ask specific questions and patience:

I am gonna let you two figure this one out while both are very useful for some coding and giving my community a good one with these handy tips.

[InJesterr]

I do have it, will test it tomorrow when I am on PC again was about to head to bed

Nvm changed my mind will do some partying but def will test tomorrow

[PunkPink]

I will be happy to share the codes I will create for this purpose and present the resources I will use to achieve the goal of this project, and I can also share the processed data with the community.

As for the procedure, I will try to detail it below:

Step 1:

Locate the location of the information.

Step 1 description:

We will explore the sites mentioned above in search of all the sub-links that may be within them. This will not be done by GPT with a prompt like the one you did in your chat, but will be carried out with specific tools or through code and the other part manually.

The first part will be automated, but as you will see in the following example, it will require some manual intervention in some cases:

For example:

(Decentraland DAO | Instagram | Linktree) includes the following URL (among others):

and, (Notion – The all-in-one workspace for your notes, tasks, wikis, and databases.) includes the following URL (among others):

(Notion – The all-in-one workspace for your notes, tasks, wikis, and databases.)

You will see that the latter includes a series of slides which are stored at: I. Fundamentals & Basic Concepts - Google Slides

There will be some URLs that, despite being outbound, contain official and relevant information from the platform, in this case manual intervention comes into play.

Step 2:

Extract contextual information

Step 2 description:

In this step, all titles, subtitles, paragraphs, etc., will be extracted. This can be done by programming a Python script to extract what is contained in the p, h1, h2, h3, and other tags.

Of course, in many cases, the information may not be contained within these tags, making manual intervention necessary, as might be the case with this URL: I. Fundamentals & Basic Concepts - Google Slides

These data, like many others that are in certain corners difficult to access by ChatGPT, will be collected (in some cases manually) to result in a .txt file (text file).

Step 3:

Extract graphical information

Step 3 description:

Images that present relevant documentation will be extracted. Understandably, this will be almost entirely a manual selection process due to the nature of this material, since, although GPT can tell you what objects and text are inside each image, it may not be able to determine which image is more relevant for the purpose of this ChatGPT:

See example:

Origin of the 1st image: https://github.com/decentraland/architecture/blob/main/docs/architecture.png
Origin of the 2nd image: I. Fundamentals & Basic Concepts - Google Slides

Step 4:

Selection of contextual material.

Step 4 description:

At this point, everything that is relevant will be selected, and redundancies, elements that may interfere with the model’s response for interpretation and eventual content, such as calendars, events, wearables, or data that may have slipped through, will be eliminated.

Step 5:

Processing of contextual material.

Step 5 description:

For this, a .csv or .txt file will be composed which will contain all the information in a structured manner to facilitate interpretation by ChatGPT to ensure that the model has access to all the information and avoid failed responses.

I hope this helps you to understand more closely the procedure that is intended to be followed.

[PunkPink]

I will answer your questions directly with Yes or No to finally resolve your doubts:

@SugarClub

1. To compile a few text documents and images from the resources you mentioned and feed it into the ChatGPT builder in add documents ?

No, the compilation of all documentation will be done outside of ChatGPT.

@SugarClub

2. To create a simple python script and add it into “actions” of the ChatGPT builder to parse for basic html headings and paragraphs?

No, the script or tools used will run outside of ChatGPT.

@SugarClub

3. To add URLs like I did ?

No, the information will be added using .csv or .txt files after it has been processed, while the selected images will be included in the traditional way as anyone could do it.

@SugarClub

Is there anything else you will do ?

Yes, any improvement that can be made to DCL-Docs during the project that has not been contemplated in this proposal will be implemented as long as it improves the usability of the mentioned tool.

@SugarClub

Compiling some pdfs and images from the resources and making a small python script could probably be done using ChatGPT itself without to much trouble

Yes, but as I have emphasized before, this proposal goes beyond simply compiling some PDFs, it is proposed to do so with all the information for a complete improvement of the tool with processed information, as for the origin of the script code it is indifferent as long as it works correctly .

@SugarClub

it would be much more interesting to ask for funding to make a specialized GPT that can build scenes using Decentraland APIs in a automated manner …: like make a scene with a small house and a tree and boom a scene is built in low poly .

Yes, but it would be a project that has no relation to what is proposed here and would have a much higher cost.

@SugarClub

In any case If this props passes you will provide all the information for others to reuse .

Yes, in fact everything used will be published on Github for the benefit of the community, this will allow the community to make improvements, updates and other changes that are considered appropriate independently.

Cheers.

[dao]

ChatGPT DCL-Docs Improvement

This proposal is now in status: PASSED.

Voting Results:

• Yes 85% 5,167,166 VP (90 votes)
• No 14% 863,493 VP (6 votes)
• Abstain 1% 1,484 VP (4 votes)

[dao]

ChatGPT DCL-Docs Improvement

This proposal has been ENACTED by a DAO Committee Member (0xfb1afa4dc069ffb47b19dbee196045d508fcd5a2)

Vesting Contract Addresses: 0x8847522eb99d35798c2a14e70905369053f51b51

[PunkPink]

Dear DAO Committee (@HPrivakos @Tobik @rizk),

We would like to request an extension for the vesting contract on “ChatGPT DCL-Docs Improvement” grant.

After our discussion with GSS, it has been determined that we require an additional month to finalize. Could you re-create the contract with a 2 month duration instead of 1?

Thank you very much.

[palewin]

Hello DAO Committee @HPrivakos @rizk @Tobik. Quick reminder: could you re-create the contract as requested? It has to be before January 12th. Thanks!
cc: @Zino

[dao]

ChatGPT DCL-Docs Improvement

This proposal has been ENACTED by a DAO Committee Member (0xfb1afa4dc069ffb47b19dbee196045d508fcd5a2)

Vesting Contract Addresses: 0x8847522eb99d35798c2a14e70905369053f51b51,0xf68BEa01EB62a2348e91E74CF79a8F994C9519a6

[Zino]

FYI: the change was made today.