List your company

Mathematical Reasoning on OlympiadBench 675 (Accuracy)

56.85AccuracyDAPO w/ Forking Tokens
28.468435.836743.20550.5733Sep 20, 2025

Evaluation results

MethodLinks
DAPO w/ Forking TokensBackbone=Qwen3-14B2025.0956.85
HAPOBackbone=Qwen3-14B2025.0956.43
Entropy EnvBackbone=Qwen3-14B2025.0954.12
ArcherBackbone=Qwen3-14B2025.0953.19
EDGE-GRPOBackbone=Qwen3-14B2025.0952.45
Vanilla DAPOBackbone=Qwen3-14B2025.0952.28
Vanilla GRPOBackbone=Qwen3-14B2025.0950.14
DAPO w/ Forking TokensBackbone=Qwen3-8B2025.0946.27
HAPOBackbone=Qwen3-8B2025.0945.26
Vanilla DAPOBackbone=Qwen3-8B2025.0944.87
EDGE-GRPOBackbone=Qwen3-8B2025.0943.35
Vanilla GRPOBackbone=Qwen3-8B2025.0942.93
Entropy EnvBackbone=Qwen3-8B2025.0942.88
ArcherBackbone=Qwen3-8B2025.0942.69
HAPOBackbone=Qwen2.5-Math-7B2025.0936.94
DAPO w/ Forking TokensBackbone=Qwen2.5-Math-7B2025.0935.52
Vanilla DAPOBackbone=Qwen2.5-Math-7B2025.0934.35
HAPOBackbone=Qwen2.5-Math-1.5B2025.0933.86
Entropy EnvBackbone=Qwen2.5-Math-7B2025.0933.68
ArcherBackbone=Qwen2.5-Math-7B2025.0933.04
DAPO w/ Forking TokensBackbone=Qwen2.5-Math-1.5B2025.0932.81
Vanilla GRPOBackbone=Qwen2.5-Math-7B2025.0932.76
EDGE-GRPOBackbone=Qwen2.5-Math-7B2025.0932.43
Vanilla DAPOBackbone=Qwen2.5-Math-1.5B2025.0931.47
EDGE-GRPOBackbone=Qwen2.5-Math-1.5B2025.0930.67
ArcherBackbone=Qwen2.5-Math-1.5B2025.0929.98
Vanilla GRPOBackbone=Qwen2.5-Math-1.5B2025.0929.84
Entropy EnvBackbone=Qwen2.5-Math-1.5B2025.0929.56