Faster assisted generation support for Intel Gaudi

In three linesHugging Face adds assisted generation support for Intel Gaudi, accelerating language model inference. The technique uses a smaller, faster model to generate candidate tokens validated by the main model, reducing overall latency.

Summary generated by Claude — human-verified