Method truncate_sequence

examples/server/server.py:14861–14878

(
        self,
        seq_id: int,
        keep_len: int,
        *,
        truncate_draft: bool = True,
    )

            request.on_done(result)

    def truncate_sequence(
        self,
        seq_id: int,
        keep_len: int,
        *,
        truncate_draft: bool = True,
    ) -> None:
        current_len = self.radix_trie.length(seq_id)
        if current_len <= keep_len:
            return
        keep_pos = self.sequence_history.position_length_for_prefix(seq_id, keep_len)
        if not llama_cpp.llama_memory_seq_rm(self.model.mem, seq_id, keep_pos, -1):
            raise RuntimeError(
                f"failed to truncate model sequence {seq_id} at position {keep_pos}"
            )
        if truncate_draft:
            self.model.truncate_draft_sequence(seq_id, keep_pos)
        self.truncate_sequence_metadata(seq_id, current_len, keep_len)

    def copy_sequence_state(
        self,