GGUF version of Locutusque/TinyMistral-248M-v2-Instruct.
do_sample: true
temperature: 0.1
top_p: 0.14
top_k: 12
repetition_penalty: 1.1
<|im_start|>user\n{user message}<|im_end|>\n<|im_start|>assistant\n{assistant message}<|endoftext|>
LocalMaxxingLD_LIBRARY_PATH=/home/beta1/qwen38-release/tools/llama-b10442-cuda /home/beta1/qwen38-release/tools/llama-b10442-cuda/llama-bench -m /mnt/pf/pf-deep2/TinyMistral-248M-v2-Instruct-GGUF/TinyMistral-248M-v2-Instruct.Q4_K_M.gguf -p 512 -n 128 -ngl 99 -fa 1在 LocalMaxxing 上查看测试
来自 LocalMaxxing 的 1 次社区测试（1 种配置）的中位数。
自己运行：配置指南