IndexTeam/Index-1.9B-Character

Index-1.9B

Model Introduction

The Index-1.9B-Character model is a role-playing model with 1.9 billion parameters in the large language model series independently developed by the Index team. Users can customize the few shots role by uploading the role dialogue materials that meet the requirements.

For usage and related codes, please see Index-1.9B/roleplay

For more details, see our GitHub and Index-1.9B Technical Report

The Index-1.9B-Character model is evaluated using the authoritative benchmark CharacterEval in the role-playing field. The evaluation set evaluates multiple granularities from the perspectives of character consistency, dialogue ability, and role-playing attractiveness. Our 1.9b model ranks ninth in overall average score, significantly better than other models of the same level.

Model	Average score	character consistency	dialogue ability	role-playing attractiveness
CharacterYuyan-RLHF	3.36	2.856	4.037	3.188
CharacterYuyan	3.309	2.88	3.931	3.117
BC-NPC-Turbo	3.169	2.764	3.798	2.946
MiniMax	3.135	2.718	3.784	2.902
Baichuan2-13B	3.129	2.701	3.795	2.89
InternLM-20B	3.124	2.715	3.745	2.911
Baichuan2-7B	3.11	2.7	3.757	2.874
Index-1.9B-character	3.109	2.676	3.789	2.862
InternLM-7B	3.034	2.62	3.698	2.784
XVERSE-13B	3.022	2.632	3.605	2.83
Qwen-14B	3.016	2.649	3.542	2.858
GPT-4	3.006	2.697	3.448	2.873
Xingchen	2.991	2.595	3.646	2.732
XVERSE-7B	2.963	2.564	3.554	2.772
CharacterGLM	2.937	2.493	3.623	2.695
ChatGLM3-6B	2.898	2.556	3.399	2.739
Qwen-7B	2.849	2.54	3.327	2.679
GPT-3.5	2.381	2.101	2.749	2.293