Ask ChatGPT /ask

[0x4007]

There have been several instances (including with myself) where I would answer a question presented in a pull request review, or in an issue conversation, by asking ChatGPT and pasting in the results.

It could be very nice to see what exactly the original prompt was inside of the conversation for full context. Imagine if we can simply handle this by using a /ask command? Any of the words following the command would be passed into GPT4.

On one hand, it feels a bit extraneous as a feature. On the other hand, we do plan to lean in pretty heavily into the AI powered features for version one of the bot, so I feel that this idea is not totally off course.

I got this from GPT
@pavlovcik

One possible solution to this problem is to modify the foundry test scenario so that it checks for a range of acceptable facet numbers, instead of a fixed number. For example, instead of testing for an exact number of facets, the test could be modified to check that the number of facets falls within a certain range, such as between a minimum and maximum value. This would allow for some flexibility in the number of facets that are acceptable and would be less likely to break with each pull request.

Another possible solution is to automate the process of updating the foundry test scenario with each pull request. This could be done using a script or tool that automatically reads the number of facets from the 3D model or other source and updates the test scenario accordingly. This would ensure that the test scenario always matches the actual number of facets in the model and would reduce the likelihood of errors or discrepancies.

It may also be possible to use a combination of these approaches, such as setting a range of acceptable facet numbers and automating the process of updating the test scenario with each pull request. This would provide both flexibility and consistency in the testing process.

Originally posted by @FibrinLab in ubiquity/ubiquity-dollar#629 (comment)

[Keyrxng]

/start

[ubiquibot]

Deadline	Wed, 23 Aug 2023 17:36:41 UTC
Registered Wallet	0xAe5D1F192013db889b1e2115A370aB133f359765

Tips:

• Use /wallet 0x0000...0000 if you want to update your registered payment wallet address @user.
• Be sure to open a draft pull request as soon as possible to communicate updates on your progress.
• Be sure to provide timely updates to us when requested, or you will be automatically unassigned from the bounty.

[Keyrxng]

Take a look here

[0x4007]

https://api.github.com/repos/ubiquity/ubiquibot/issues/291/comments

Scrubbed:

[
  {
    "id": 1690282995,
    "user": {
      "login": "Keyrxng",
      "id": 106303466,
      "type": "User"
    },
    "created_at": "2023-08-23T16:36:38Z",
    "updated_at": "2023-08-23T16:36:38Z",
    "body": "/start"
  },
  {
    "id": 1690283105,
    "user": {
      "login": "ubiquibot[bot]",
      "id": 113181824,
      "type": "Bot"
    },
    "created_at": "2023-08-23T16:36:43Z",
    "updated_at": "2023-08-23T16:36:43Z",
    "body": "\n<code>\n\n  <table>\n  <tr>\n    <td></td>\n    <td></td>\n  </tr>\n  <tr>\n    <td>Deadline</td>\n    <td>Wed, 23 Aug 2023 17:36:41 UTC</td>\n  </tr>\n  <tr>\n    <td>Registered Wallet</td>\n    <td>0xAe5D1F192013db889b1e2115A370aB133f359765</td>\n  </tr>\n  \n  \n  \n  </table>\n</code><h6>Tips:</h6>\n    <ul>\n    <li>Use <code>/wallet 0x0000...0000</code> if you want to update your registered payment wallet address @user.</li>\n    <li>Be sure to open a draft pull request as soon as possible to communicate updates on your progress.</li>\n    <li>Be sure to provide timely updates to us when requested, or you will be automatically unassigned from the bounty.</li>\n    <ul>"
  },
  {
    "id": 1690302671,
    "user": {
      "login": "Keyrxng",
      "id": 106303466,
      "type": "User"
    },
    "created_at": "2023-08-23T16:50:15Z",
    "updated_at": "2023-08-23T16:50:15Z",
    "body": "Take a look [here](https://github.com/Keyrxng/didactic-octo-train/issues/8)"
  }
]

[Keyrxng]

I don't think there's default config anymore, I had a hard time using my repo after some recent update..

It kept telling me to remove some data in config that's not needed and add some new ones, also renamed some before it started working again

It done the same for myself but looking at your repo config you only have two here.

Where is it reading the rest from for you?

^^^^^

Scratch that, your dev branch is 90 commits behind just noticed and not sure which is your most recent branch to check myself

[Keyrxng]

ahhh okay maybe I'll have to do the same then if that's the only way things are working at the moment but not ideal

[seprintour]

I'm going to test with my repo now, besides I think #796 needs to be fixed. It's not easily noticeable but new updates will not reflect with this bug

[Keyrxng]

I spotted whilefoo raise this on TG at the time and tried to reproduce that and could not, to this day still haven't and i've done a fresh install multiple times and again like 10 mins ago so #796 doesn't affect me somehow

[ubiquibot]

Do you have any updates @Keyrxng? If you would like to release the bounty back to the DevPool, please comment /stop
Last activity time: Sat Sep 23 2023 18:19:11 GMT+0000 (Coordinated Universal Time)

[Keyrxng]

/start

[ubiquibot]

Deadline	Mon, 09 Oct 2023 18:16:41 UTC
Registered Wallet	0xAe5D1F192013db889b1e2115A370aB133f359765

Tips:

• Use /wallet 0x0000...0000 if you want to update your registered payment wallet address @user.
• Be sure to open a draft pull request as soon as possible to communicate updates on your progress.
• Be sure to provide timely updates to us when requested, or you will be automatically unassigned from the bounty.

[Keyrxng]

/start

[ubiquibot]

Skipping /start since the issue is already assigned

[Keyrxng]

• /ask ubq-testing/sandbox#1 (comment)

It's working on my org repo now that I've gotten around the default config still not working

All I've done is pass in 4000 via the private settings repo with the path configRepo/.github/config.yml

Is this still relevant? Busy week there but didn't think it was resolved

Using langchain we can pass -1 for token count which just passes the remaining tokens in as the requested response, just a little fyi

this.llm = new OpenAI({
            openAIApiKey: this.apiKey,
            modelName: 'gpt-3.5-turbo-16k',
            maxTokens: -1,
        })

This would be an easy and simple fix for our token problems

I'm still getting the no such file error for \lib\assets\images\pmg.png']. I've created /lib/assets/pmg.png and still noy joy, it's exiting the process after responding for me. Any fix on this?

[Keyrxng]

I'm assuming that you've probably used an api key that is used for other things but @pavlovcik by chance have you created a key that we can check the usage to see whether the /ask call is being made when called from this issue?

I cannot reproduce the non-response I get errors or it responds to me as it should

[Keyrxng]

• b417336

Introduced `tokenLimit: openAITokenLimit || 0,`` which will result in it failing if the tokenLimit is undefined as 0 is an invalid value for max_tokens. So we'll have to define a reason value as rndquu requested before, my recommendation is probably about 60/40 for the size of the issues being parsed and linked in this org

[Keyrxng]

It's probably not the case but I'm sure I said it before but parallel /asks on different repos would fail for me on at least one them. This would imply every time it's been called was at the same time as someone else calling it which is absurd but still something to consider for down the road I think

[Keyrxng]

@pavlovcik @rndquu bump

bumping as I feel I've dragged this out far longer than acceptable with being MIA last week, hoping to sort the 3 out this week asap

[ubiquibot]

Do you have any updates @Keyrxng? If you would like to release the bounty back to the DevPool, please comment /stop
Last activity time: Wed Oct 04 2023 09:36:18 GMT+0000 (Coordinated Universal Time)

[rndquu]

The /ask command is working fine in the latest development branch.

The thing is that right now the bot's production build is set to the 1st of September which doesn't have the ask command codebase so we can't use it in production until we switch to the latest bot's build.

Anyway this issue can be closed as completed because all we need to do is:

1. Switch to the latest bot's build
2. Set openai bot params (already set)

[Keyrxng]

Yes @rndquu, love to hear it! So everything is all good with this, so I can finish the rest of the pr's that rely on and/use the same functionality. I knew there wasn't anything on my end I done everything I could think of to debug and resolve this lmao was rather shitting it truth be told so I'm glad you have put this to bed

That "tokenLimit || 0" needs to be updated as that'll cause headaches but all good otherwise

[ubiquibot]

Do you have any updates @Keyrxng? If you would like to release the bounty back to the DevPool, please comment /stop
Last activity time: Tue Oct 10 2023 14:07:18 GMT+0000 (Coordinated Universal Time)

[Keyrxng]

Everything is working as it should be

removed my assignment to avoid any more bot updates