List your company

General Reasoning on GPQA (Avg@k, Pass@k)

35.66Avg@kSFT
26.081628.568331.05533.5417May 8, 2026

Evaluation results

MethodLinks
SFTBackbone=Qwen2.5-7B-Instruct2026.0535.6665.63
IWBackbone=Qwen2.5-7B-Instruct2026.0534.8264.29
IWBackbone=Llama-3.1-8B-Instruct2026.0534.3269.2
RPSFTBackbone=Qwen2.5-7B-Instruct2026.0534.2165.18
RPSFTBackbone=Llama-3.1-8B-Instruct2026.0533.4868.08
BaseBackbone=Qwen2.5-7B-Instruct2026.0532.7659.82
SFTBackbone=Llama-3.1-8B-Instruct2026.0532.765.4
DFTBackbone=Qwen2.5-7B-Instruct2026.0530.4758.04
IWBackbone=Qwen2.5-3B-Instruct2026.0529.4165.4
RPSFTBackbone=Qwen2.5-3B-Instruct2026.0528.1362.05
SFTBackbone=Qwen2.5-3B-Instruct2026.0527.7361.61
BaseBackbone=Llama-3.1-8B-Instruct2026.0527.6261.38
BaseBackbone=Qwen2.5-3B-Instruct2026.0527.1854.46
DFTBackbone=Llama-3.1-8B-Instruct2026.0526.7957.14
DFTBackbone=Qwen2.5-3B-Instruct2026.0526.4555.13