SOTA Science Question Answering on ARC-c (test) and PapersWithCode

91.3Accuracy

Qwen3-8B + SFT

Updated 22d ago

Evaluation Results

Method
Qwen3-8B + SFT 2026.05	91.3	-
Qwen3-8B + SFT + WeMask(TF) 2026.05	91.3	-
Qwen3-8B + WeMask(SFT) 2026.05	91.3	-
Trinity Large (MoE) 2026.02	90	-
Qwen3 8B 2026.02	88	-
Qwen3 4B 2026.02	84	-
Qwen-1.5 14B (Teacher) 2024.07	80.59	-
Llama3.1-8B-Instruct + WeMask(SFT) 2026.05	79.69	-
Llama3.1-8B-Instruct + SFT 2026.05	79.52	-
Llama3.1-8B-Instruct + SFT + WeMask(TF) 2026.05	79.35	-
Llama3.1-8B-Instruct 2026.05	74.49	-
Datology 8B 2026.02	72	-
Llama-3.1 8B 2026.02	67	-
Granite-4.0 Micro 2026.02	67	-
SmolLM3 3B 2026.02	65	-
LFM2.5 1.2B 2026.02	65	-
Datology 3B 2026.02	63	-
FLAN-137B 2026.03	61.7	-
Imagine-DeBERTa-v3-L (Retrieval) 2026.03	59.5	-
Imagine-DeBERTa-v3-L 2026.03	59.2	-
Llama-3.2 3B 2026.02	58	-
Imagine-DeBERTa-v3-L 2026.03	56.2	-
Qwen-1.5 1.8B + DDK 2024.07	55.03	-
Qwen-1.5 1.8B + TED 2024.07	55	-
Qwen-1.5 1.8B + KD 2024.07	54.56	-
Qwen-1.5 1.8B + MiniLLM 2024.07	53.92	-
GPT-3 2021.12	53.2	-
CAR-DeBERTa-v3-L 2026.03	53.2	-
Qwen-1.5 1.8B + CPT & DoReMi 2024.07	52.11	-
GLaM 2021.12	52	-
GPT-3 2021.12	51.5	-
GPT-3 2021.12	51.4	-
Qwen-1.5 1.8B + CPT 2024.07	51.03	-
Qwen-1.5 1.8B (Student) 2024.07	50.31	-
GLaM 2021.12	50.3	-
GLaM 2021.12	48	-
Imagine-RoBERTa-L 2026.03	39.1	-
CAR-RoBERTa-L 2026.03	36.5	-
Z-LaVI (BART-L) 2026.03	36.5	-
Imagine-GPT-2-L 2026.03	35.1	-
GPT-J-6B 2026.03	34.8	-
OPT-30B 2026.03	34.8	-
Z-LaVI (OPT-30B) 2026.03	34.1	-
Z-LaVI (RoBERTa-L) 2026.03	33.4	-
Llama-3.2 1B 2026.02	33	-
GPT-Neo-2.7B 2026.03	31.8	-
SMLM 2026.03	28.4	-
Qwen3-8B 2026.05	22.78	-