emfhal/ gpt-oss:20b-cpu-optimized

434 2 months ago

CPU-optimized gpt-oss:20b for Kubernetes and VM environments. Runs without GPU, tuned for predictable CPU inference, balanced quality, and reduced memory usage.

tools thinking
... /
params
620008dda6d9 · 111B
{
"num_batch": 512,
"num_ctx": 4096,
"num_gpu": 0,
"num_thread": 8,
"repeat_penalty": 1.1,
"temperature": 0.7,
"top_p": 0.9
}