I understand the feeling - LLMs have seen the source code for other projects. but would you not have read other engine’s code to understand the problem space and existing solutions too?
The architecture is different, the language is different, implementation choices are all mine. It was not developed by autonomous agents. I don’t think you’ll find any code that is a duplicate of other engines* - happy to be proven wrong here.
* The regexp and unicode modules are taken from QuickJS and have the appropriate licenses.
I understand the feeling - LLMs have seen the source code for other projects. but would you not have read other engine’s code to understand the problem space and existing solutions too?
The architecture is different, the language is different, implementation choices are all mine. It was not developed by autonomous agents. I don’t think you’ll find any code that is a duplicate of other engines* - happy to be proven wrong here.
* The regexp and unicode modules are taken from QuickJS and have the appropriate licenses.
Please explain what you mean by that.
whichever LLM GP is using, there's undoubtedly mutliple JS engines' source code in the training data, making it not really "from scratch".