Release 1.110 notes seems to have incorrect TTS tokenizer link

[alex-ie]

https://github.com/LostRuins/koboldcpp/releases/tag/v1.110

https://huggingface.co/koboldcpp/tts/blob/main/qwen3-tts-tokenizer-q8_0.gguf
for
https://huggingface.co/koboldcpp/tts/resolve/main/Qwen3-TTS-12Hz-1.7B-Base-q8_0.gguf

See blob in the address, it downloads 80kb file. Also even if blob part is fixed, will it work? it's for qwen3-tts-tokenizer-q8_0.gguf
model.

P.S. I've tried to find it elsewhere https://huggingface.co/search/full-text?q=+Qwen3-TTS-12Hz-1.7B-Base-q8_0.gguf
but "No result found" - not even present https://huggingface.co/koboldcpp/tts/resolve/main/Qwen3-TTS-12Hz-1.7B-Base-q8_0.gguf.
Why? I guess it looks only in text files, not for files as Qwen3-TTS-12Hz-1.7B-Base finds many in readme.md-s.

Maybe it will be beneficial for the project to add file names in https://huggingface.co/koboldcpp/tts/resolve/main/README.md?

[LostRuins]

Thank you for pointing it out, i have fixed the URL. The correct link is https://huggingface.co/koboldcpp/tts/resolve/main/qwen3-tts-tokenizer-q8_0.gguf

Yes, it should work.

[alex-ie]

Yes, it should work.

@LostRuins Since starting to use kcpp about a month ago I got the impression a tokenizer file can be used only for same model for same number of parameters.

But qwen3-tts-0.6b-q8_0.gguf (1.3Gb) and Qwen3-TTS-12Hz-1.7B-Base-q8_0.gguf (2.4Gb) are what from name looks to have 0.6 and 1.7B. So different number of parameters are sometimes OK? How to know that? Only trying to run?

[LostRuins]

The two models happen to work on the same tokenizer so that's fine, but its not always the case. You can't assume it unless its stated