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

Temporal Question Answering on TimeQA Hard

77.7EM

DeepSeek-V3-AdapTime

Updated 1mo ago

Evaluation Results

Method	Links
DeepSeek-V3-AdapTime 2026.04		77.7	-	-	79.2
DeepSeek-V3-Step-back 2026.04		76.4	-	-	77.9
DeepSeek-V3-Self-refinement 2026.04		76.4	-	-	78.4
DeepSeek-V3-CoT 2026.04		75.6	-	-	77
DeepSeek-V3-ICL 2026.04		68.8	-	-	71.6
Qwen-3-8B-AdapTime 2026.04		66.5	-	-	68.2
TG-LLM 2026.04		63.1	-	-	66.4
Qwen-3-8B-CoT 2026.04		62.9	-	-	64.7
REMEMO-large 2026.04		60.5	-	-	69.3
T5-large 2026.04		59.5	-	-	68.1
REMEMO-base 2026.04		58.2	-	-	67.3
Qwen-3-8B-ICL 2026.04		56.9	-	-	60.1
T5-base 2026.04		55.6	-	-	64.1
GPT-4 2026.04		54.6	-	-	57.1
T5-L-FiD-PIT 2023.11		52.7	47.3	49.8	61
T5-L-FiD 2023.11		50.5	45.1	47.6	59.8
T5-B-FiD-PIT 2023.11		46	41.1	43.3	54.7
T5-B-FiD 2023.11		44.3	39.4	41.7	53.2
QAaP 2026.04		39.6	-	-	49.3
T5-B-PIT 2023.11		39	34.2	36.4	48.4
T5-B 2023.11		37.3	32.9	34.9	46.8
LLaMA-3-8B-AdapTime 2026.04		33.3	-	-	38.6
LLaMA-3-8B-CoT 2026.04		31.6	-	-	33.8
T5-B-FiD 2023.11		10.3	-	-	19.7
LLaMA-3-8B-ICL 2026.04		1.7	-	-	3.6