Prompt Compression on MeetingBank (test) benchmark leaderboard · SOTA2 Research

Prompt Compression on MeetingBank (test)

21.5BLEU ScoreResults with Original Prompts

Evaluation Results

Method
Results with Original Prompts2024.09	21.5	55.19	33.03	42.9	91.12
TACO-RL2024.09	21.35	55.34	31.88	42.17	90.95
TACO-RL2024.09	19.36	53.67	29.54	40.01	90.54
LLMLingua-2 - MeetingBank2024.09	18.68	54.2	29.45	40.14	90.69
TACO-RL2024.09	17.61	52.33	27.84	38.57	90.26
LLMLingua-2 - Wikitext2024.09	16.71	52.58	27.73	39.05	90.47
TACO-RL2024.09	15.85	50.56	26.04	36.81	89.96
LLMLingua-2 - MeetingBank2024.09	15.11	51.67	25.6	37.18	90.17
TACO-RL2024.09	14.25	48.6	24.51	35.08	89.68
Selective Context2024.09	12.94	46.3	24.41	34.56	89.66
LLMLingua-2 - Wikitext2024.09	12.93	49.38	23.32	35.34	89.79
LLMLingua-2 - MeetingBank2024.09	12.8	49.4	22.77	34.77	89.78
LLMLingua-2 - MeetingBank2024.09	11.13	47.5	21.01	33.25	89.44
LLMLingua-2 - Wikitext2024.09	10.98	47.34	20.68	33.06	89.34
LLMLingua-2 - MeetingBank2024.09	9.8	45.82	19.19	31.64	89.12
LLMLingua-2 - Wikitext2024.09	9.51	45.33	18.71	31.27	88.9
Selective Context2024.09	8.8	40.22	19.28	29.44	88.67
LLMLingua-2 - Wikitext2024.09	8.55	43.44	17.06	29.67	88.53
Selective Context2024.09	6.37	36.22	15.97	26.07	88
LLMLingua2024.09	5.9	38.22	14.02	26.49	87.65
Selective Context2024.09	4.82	33.15	13.55	23.74	87.49
Selective Context2024.09	4.06	31.34	12.21	22.31	87.12
LLMLingua2024.09	3.98	32.62	10.58	22.58	86.52
LLMLingua2024.09	3.51	30.98	9.64	21.33	86.2
LLMLingua2024.09	3.42	30.53	9.62	21.15	86.14
LLMLingua2024.09	3.19	29.85	9.47	20.75	86.07