Hey, I am a pro user and I have an issue with all models during medium-to-large composter or chat conversation sessions. The model generates code and then starts to apply it, but the process gets automatically canceled. This happens with any model when working with large files (3000–5000 lines). changes and code not that the code is huge, but it fails to apply simple changes. It doesn’t work with o1 either – the model generates code, but then it gets canceled. I have lost many fast tokens and my O1 quota. Please fix this or provide a reason why it happens.
Why are your files 3000-5000 lines? Even if the model was better, AI is trained on best practices, which are modular, so its always going to have an easier time / create better solutions if you don’t fight it.
So problem is the too much information and model automatically cancelled apply?
Not necessarily too much information. If you have 5000 lines of code in one module, there’s probably bits and pieces connecting through it at all different levels. This makes it very hard for the AI to figure out what’s going on exactly. Look up SOLID design principles for breaking apart large blocks of code into smaller more manageable files. The AI does best when file sizes are 500 likes are smaller. Its a lot easier for it to jump around 10 files with 500 lines of code than 1 file with 5000.