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

Multi-hop Reasoning on HotpotQA (Accuracy, F1)

59.64Accuracy

ReLAR

Updated 1mo ago

Evaluation Results

Method	Links
ReLAR 2026.06		59.64	76.15
ReLAR 2026.06		57.5	75.23
Llama-3-8B-Instruct 2026.06		56.47	71.83
Med42-Llama3-8B 2026.06		54.01	69.82
Qwen2.5-7B 2026.06		52.64	66.83
Mistral-7B-Instruct 2026.06		50.83	64.47
Qwen2.5-Med-7B 2026.06		50.18	64.15
Llama-3-8B-Instruct 2026.06		48.83	63.47
Mistral-7B-v0.3 2026.06		47.82	61.35
MedGemma-4B 2026.06		46.91	60.05
Med42-Llama3-8B 2026.06		46.23	61.59
Qwen2.5-7B 2026.06		44.83	58.47
Gemma-7B 2026.06		42.64	56.83
Qwen2.5-Med-7B 2026.06		41.81	55.13
Mistral-7B-Instruct 2026.06		41.47	55.28
LLaMA-2-7B 2026.06		39.82	52.63
Med42-Mistral-7B 2026.06		39.64	53.47
MedGemma-4B 2026.06		39.48	53.29
Mistral-7B-v0.3 2026.06		35.63	48.74
Med42-Mistral-7B 2026.06		32.47	45.83
Gemma-7B 2026.06		31.82	43.47
Falcon-7B 2026.06		30.84	41.64
LLaMA-2-7B 2026.06		27.34	38.47
Falcon-7B 2026.06		22.47	31.83