Aiming at the ambiguity of the decision-making information for multi-attribute large-group and difference of the attribute, the interval 2-tuple linguistic large-group decision method based on the attribute of multi-granularity is proposed. Considering the preference of decision-makers, the large group is clustered and classified to form aggregation structure. According to the decision-makers’ different contribution to the existing clusters’ consistency and different contribution of different cluster, the two-layer weight model is presented. Then, the possible degree formula for expansion is used to confirm the decision-makers’ weighted value of all clusters, and the fuzzy entropy is refused to confirm the weighted value of different cluster. Furthermore, the fuzzy relative entropy and technique for order preference by similarity to an ideal solution(TOPSIS) are used to sort the schemes. Finally, the case analysis and comparison verify the effectiveness and applicability of the proposed method.