YTan2000

/

Qwen3.5-27B-TQ3_1S

like 37

still compatible with this

this isn't working on my llama.cpp:

[==================================================] 100% (13232 MB / 13232 MB)
gguf_init_from_file_impl: tensor 'token_embd.weight' has invalid ggml type 44 (NONE)
gguf_init_from_file_impl: failed to read tensor info

Please make sure you use https://github.com/turbo-tan/llama.cpp-tq3 , main branch. Type 44 and 46 are defined in there. It is a fork from llama.cpp, feel free to contribute.

added mmproj for all 3. Suggest to use 4S.