Have you fixed or conquered this obstacle
I may be regretting for not paying my ChatGPT plus subscription this month. I am using the ChatGPT-3.5 from the playground until I can pay the ChatGPT plus next week. Nonetheless I would like to bring the attention of the community on a problem that I was also experiencing when using ChatGPT-3.5 but that I am experiencing now using the ChatGPT web site in parallel of my Playground session…
Using ChatGPT-text-davinci-002-render-sha
In parallel of prompting the Playground ChatGPT-3.5 I decided to ask for help to the free ChatGPT from the official ChatGPT web site which for me is currently ChatGPT-text-davinci-002-render-sha and I got this situation where it is impossible to use the previous context and I am unsure if it is something that can be improved in that model or if I must learn how to prompt the AI but as I mentioned in a previous thread earlier today asking too much is not better than asking too little:
An other example of a circular problem
This is the part of my conversation with ChatGPT that illustrate the issue that I am trying to explain since February without any successful solution so far…
The conversation was already started:
I STILL HAVE THE PROBLEM THE CODE WILL FOLLOW THE ERROR MESSAGE
PLEASE INGEST ERROR MESSAGE FIRST AND INFER ABOUT THE ERROR
MESSAGE FIRST PLEASE
[a long error message was included]
please now ingest the code and then infer a solution please consider each part of the code provided to you :
`in /projects/images_project/typescript/src/main.ts`
[the code was included]
`in /projects/images_project/typescript/src/path-processing/cacheFolderPath.ts`
[more code]
remember those assertions are true pleas use them to infer:
both files exists and are located at: «/projects/images_project/typescript/src/main.ts» and «/projects/images_project/typescript/src/image_processing_config.json»
Reply from the AI:
Based on the error message, it seems that the code is not able to find the file image_processing_config.json in the current directory. The error message specifically says “no such file or directory, open ‘image_processing_config.json’”.
The configFilePath variable in main.ts is set to ‘./image_processing_config.json’, which means that the code is expecting the file to be in the same directory as main.ts.
Therefore, one possible solution could be to verify that the file image_processing_config.json actually exists in the correct location (/projects/images_project/typescript/src/).
Another solution could be to use an absolute path to the config file instead of a relative path. This can be done by replacing the configFilePath variable with the full path to the file, like this:
[typescript] [Copy code]
// [Copy code button]
const configFilePath = '/projects/images_project/typescript/src/image_processing_config.json';
This will ensure that the code always looks for the config file in the correct location, regardless of the current working directory.
When the groundhog day start you do not initially know
this will now be compatible with the code provided earlier look:
const configFilePath = join(__dirname, normalPath);
please tell me how to correct the code you revived earlier
Reply from the AI:
Yes, you are correct. To fix the error, you need to modify the readConfigFile function to use configFilePath instead of normalPath when calling readFileSync and also remove the process.chdir line. Here’s the corrected code for readConfigFile:
[arduino] [Copy code]
export function readConfigFile(filePath: string): ImageProcessingConfig {
const normalPath = normalize(filePath);
const configFilePath = join(__dirname, normalPath);
const configFile = readFileSync(configFilePath, 'utf-8');
const config = JSON.parse(configFile) as ImageProcessingConfig;
return config;
}
This should resolve the error you were encountering.
No it is not arduino code
I mentioned that:
« no it would now I want to use the absolute path »
Reply from the AI going back fully circular:
Sure, to use an absolute path instead of a relative path, you can replace the configFilePath variable assignment in main.ts with an absolute path like this:
[typescript] [Copy code]
import * as path from 'path';
// ...
const configFilePath = path.join(__dirname, 'image_processing_config.json');
This should construct an absolute path by joining the directory of the current module (__dirname) with the filename image_processing_config.json.
So the problem is that I ask one thing then I ask a correction, then I receive the correction but if I ask something else bam the spark of AGI is gone and we are back the initial promblem.
My quest to find people with the same struggle
I am seeking to find anyone else who feel the same frustration, or I would like a professional prompter to tell me what I am doing wrong (aside of using ChatGPT-text-davinci-002-render-sha instead of paying for ChatGPT-4) or lastly I would like to know how can I get some feedback from OpenAI about this ongoing situation (I have had the same issues with ChatGPT-3.5 since late February).
My quest to see this behaviour investigated by OpenAI
My intuition is that it is a behaviour that should be addressed, since the model text-davinci-002-render is an older model that will soon be obsolete I understand that it will not probably be possible to fix it…
I still believe non the less that this is a behaviour that should be looked at and fixed in the flowing updated of the more recent APIs and Chat Models.
My quest to find professional prompter to overcome the situation
If someone can understand what I am trying to say and can tell me something about their experiences without doing some kind of whataboutism (what about using the Playground, Whatabout using the paid version, Whatabout learning how to prompt an AI)… I would like to hear from you guys and girls and everyone else about your take on this situation. My goal is to find constructive feedback that would help me or the rest of the community or that would help OpenAI make better products.
And as I said few paragraphs ago if anyone know a professional prompter to tell me what I am doing wrong, I believe that the AI is wrong, but I am still open to improve myself…
↑A woman, a man or anyone who is a professional prompter Please let me know waht you think!