Function createOrdinaryContinuationRequest

createOrdinaryContinuationRequest(
    input: {
        execution: ResolvedExecution;
        interval: number;
        onToken: TokenSink;
        options: GenerateOptions;
        prompt: number[];
    },
): {
    continuation: OrdinaryContinuation;
    dispose: () => void;
    onToken: TokenSink;
    plainGreedy: boolean;
    sample: (logits: MlxArray, step: number) => MlxArray;
}
Optional policy and sampler recovery for the existing ordinary driver. The cache service resumes and checkpoints the generation under the key this request names; KV layout and delayed conversion stay with the caches.
Parameters
- input: {
      execution: ResolvedExecution;
      interval: number;
      onToken: TokenSink;
      options: GenerateOptions;
      prompt: number[];
  }
  - execution: ResolvedExecution
  - interval: number
    
    Checkpoint every this many generated tokens (the cache service's checkpointEveryTokens).
  - onToken: TokenSink
  - options: GenerateOptions
  - prompt: number[]
Returns {
    continuation: OrdinaryContinuation;
    dispose: () => void;
    onToken: TokenSink;
    plainGreedy: boolean;
    sample: (logits: MlxArray, step: number) => MlxArray;
}
- Defined in packages/inference/src/execution/continuation-request.ts:12