Anonymous
7/12/2025, 11:21:53 PM
No.105885592
>>105885540
As you very well know, the settings on the gguf are taken from the original model's config. If there's a problem with those settings, you check the gguf settings (shown on load or can be extracted directly) and the original config.
The problem doesn't originate in llama.cpp, it originates on the original config files. The program just does what its told. Luckily, some dev took the time to add a runtime check for that to let you know. And some other dev added the flags you need to fix what the model makers didn't.
As you very well know, the settings on the gguf are taken from the original model's config. If there's a problem with those settings, you check the gguf settings (shown on load or can be extracted directly) and the original config.
The problem doesn't originate in llama.cpp, it originates on the original config files. The program just does what its told. Luckily, some dev took the time to add a runtime check for that to let you know. And some other dev added the flags you need to fix what the model makers didn't.