Page 1 of 1

Large Language Models that can decompile Shandalar

PostPosted: 28 Mar 2024, 15:52
by harvestman122
The programmers (like @CirothUngol) here will be interested in LLM4Deompile, which is an LLM-based decompiling AI. It's capable of decompiling such languages as C++ and so it can turn Shandalar into human-readable code. While OpenAI's GPT4 is great at this as well, LLM4Decompile is substantially better as well as being free if you set it up with a Google Collab or run it locally on your computer.

Unfortunately, I can't place external links in my post, but if you check on Hugging Face for "LLM4Decompile" you'll see a lot of different versions of this AI for free use with a Google Collab.

Re: Large Language Models that can decompile Shandalar

PostPosted: 19 Apr 2024, 22:17
by Huggybaby
Thanks, I was talking to someone about this exact subject the other day, but hadn't heard of this project.
https://github.com/heckerstone/LLM4Deco ... /README.md
https://huggingface.co/papers/2403.05286
https://huggingface.co/arise-sustech/ll ... le-6.7b-uo