Fetching the paper…

LazyLLM: Dynamic Token Pruning for Efficient Long Context LLM Inference · Around