A user says a strong AI coding model still makes basic mistakes
A Reddit user says they saw many recent posts praising Fable for fixing bugs and building things like a full game. They were first impressed by how capable it felt, but then ran into trouble while using it on a simple coding mistake. The user says Fable went in circles, broke unrelated parts of the code, and showed that it is still an LLM that can make mistakes and focus too much on one issue.
Key points
- The post says Fable has been praised for fixing bugs and building larger projects.
- The user says Fable feels more capable than earlier models.
- The user had trouble with Fable on a simple coding mistake.
- Fable reportedly repeated the same issue and broke unrelated code.
- The user sees Fable as powerful, but still an LLM that needs checking.
Quick term guide
- build
- A chosen set of in-game abilities or items a player equips for their character.
- issue
- A written task, bug, or feature request in a software project.
- Solo makers
- People who build and launch their own products or services entirely on their own.
- tools
- Outside functions an AI can use, such as search, calculation, or code running.
- checkpoints
- Saved model files that can be loaded for testing or use.
- checkpoint
- A saved AI image model file that usually carries a certain style or capability.
- tests
- Automatic checks that help confirm code works as expected.
- models
- Different AI engines that can power answers or code suggestions inside a tool.