Do we need to reuse instance?

[linonetwo]

In the example

const llama = new LLM(LLamaCpp);
const config: LLamaLoadConfig = {
  modelPath,
  enableLogging: true,
  nCtx: 1024,
  seed: 0,
  f16Kv: false,
  logitsAll: false,
  vocabOnly: false,
  useMlock: false,
  embedding: false,
  useMmap: true,
  nGpuLayers: 0,
};
await llama.load(config);
await llama.createCompletion(

do we need to store the llama variable in global, to reuse it? Will it speedup or reduce memory usage?

And I found await llama.createCompletion never ends, will this cause memory leak?

[linonetwo]

Seems rwkv.cpp will re-loading .bin from disk every time I call createCompletion