Qwen 3.8 27B KV f16 vs q8_0 are not equivalents
英文标题:Qwen 3.8 27B KV f16 vs q8_0 are not equivalents
I'm testing it since release, now with UD 3.0 in my AMD R9700 with ROCm, I always read everywhere that F16 and q8_0 for KV cache are essentially the same... well, I tested it and I can see differences. Some differences are minimal, F16 is more careful and detailed with structured and free output, th
打开原文相关 2 · 机会 0
中文
I'm testing it since release, now with UD 3.0 in my AMD R9700 with ROCm, I always read everywhere that F16 and q8_0 for KV cache are essentially the same... well, I tested it and I can see differences. Some differences are minimal, F16 is more careful and detailed with structured and free output, th
中文全文翻译正在补齐,可先查看下方英文原文。
English / 原文
I'm testing it since release, now with UD 3.0 in my AMD R9700 with ROCm, I always read everywhere that F16 and q8_0 for KV cache are essentially the same... well, I tested it and I can see differences. Some differences are minimal, F16 is more careful and detailed with structured and free output, th