Missing Qwen3-8B-Q4_0.gguf
#4
by
StellanWay
- opened
It's a legacy format, but the other unsloth repositories (Qwen3-4B-GGUF, Qwen3-14B-GGUF) have it and it's the only format supported by llama.cpp's OpenCL backend. Inference is much faster on my phone (OnePlus 13 with Qualcomm Snapdragon 8 Elite) with OpenCL.