Share your thoughts, 1 month free Claude Pro on usSee more

Question Answering on MultiFieldQA en

24.67EM

BRIEF-PRO-AUTO

Updated 1mo ago

Evaluation Results

Method	Links
BRIEF-PRO-AUTO 2025.10		24.67	54.21	11
BRIEF-PRO-HIGH 2025.10		24	52.43	29
BRIEF-PRO-MEDIUM 2025.10		24	52.23	16
BRIEF-PRO-LOW 2025.10		24	53.88	10
ProLong-8B 2025.10		22	51.44	1
FILM-7B 2025.10		22	54.34	1
Non-compression 2025.10		21.33	52.77	1
Non-compression 2025.10		21.33	54.38	1
BRIEF-PRO-HIGH 2025.10		21.33	50.7	29
BRIEF-PRO-AUTO 2025.10		20.67	51.16	11
BRIEF-PRO-MEDIUM 2025.10		20.67	51.93	16
BRIEF-PRO-LOW 2025.10		20.67	50.63	10
RECOMP (Extractive) 2025.10		18	46.43	16
BRIEF-PRO-HIGH 2025.10		18	48.53	29
RECOMP (Extractive) 2025.10		17.33	43.82	16
LongLLMLingua 2025.10		17.33	44.62	4
Rerank Top-3 2025.10		17.33	45.1	3
BRIEF-PRO-LOW 2025.10		17.33	49.75	10
LongLLMLingua 2025.10		16.67	45.87	4
Non-compression 2025.10		16.67	50.09	1
BRIEF-PRO-AUTO 2025.10		16.67	48.17	11
BRIEF-PRO-MEDIUM 2025.10		16.67	49.79	16
Rerank Top-3 2025.10		16.5	43.1	3
Llama-3.2-3B-Instruct 2025.10		14.67	42.3	16
GPT-4.1-nano 2025.10		14.67	41.95	29
LongLLMLingua 2025.10		14	44.88	4
GPT-4.1-nano 2025.10		13.33	41.44	29
Rerank Top-3 2025.10		13.33	42.67	3
Llama-3.2-3B-Instruct 2025.10		12.8	39.31	16
RECOMP (Extractive) 2025.10		12.67	43.72	16
Llama-3.2-3B-Instruct 2025.10		12.67	41.55	16
GPT-4.1-nano 2025.10		12.67	41.04	29
RECOMP (Abstractive) 2025.10		11.33	33.78	12
EXIT 2025.10		10.67	25.33	14
EXIT 2025.10		10.67	25.75	14
EXIT 2025.10		9.33	27.12	14
RECOMP (Abstractive) 2025.10		8	30.14	12
RECOMP (Abstractive) 2025.10		7.33	28.83	12