Poisoning Attack on RAG on HotpotQA (test) benchmark leaderboard · SOTA2 Research

Poisoning Attack on RAG on HotpotQA (test)

100PoisonedRAG Success Rateqwen/qwen2.5-7b-instruct

Evaluation Results

Method
qwen/qwen2.5-7b-instruct2026.03	100	93	96	1	4	100	0
qwen/qwen2-7b-instruct2026.03	100	96	89	22	6	100	0
ibm/granite-3.3-8b-instruct2026.03	100	99	98	10	7	100	0
meta/llama-4-maverick-17b-128e-instruct2026.03	100	95	100	16	5	100	0
meta/llama-3.1-8b-instruct2026.03	98	92	92	1	8	98	0
meta/llama-3.3-70b-instruct2026.03	95	94	90	17	5	100	5
openai/gpt-oss-120b2026.03	93	91	94	5	4	100	7
openai/gpt-oss-20b2026.03	90	94	85	3	6	97	7