List your company

Mathematical Reasoning on OlympiadBench (Avg@16)

50.7Avg@16HEAPA
12.7422.59532.4542.305Jan 30, 2026

Evaluation results

MethodLinks
HEAPABackbone=Qwen3-8B2026.0150.7
HEAPABackbone=Qwen3-4B2026.0146.5
GRPO (w. PS)Backbone=Qwen3-8B2026.0145.6
GRPO (w. PS)Backbone=Qwen3-4B2026.0141.1
GRPOBackbone=Qwen3-8B2026.0139.4
GRPOBackbone=Qwen3-4B2026.0133.1
HEAPABackbone=Qwen3-1.7B2026.0129.4
GRPO (w. PS)Backbone=Qwen3-1.7B2026.0128.2
GRPOBackbone=Qwen3-1.7B2026.0126.5
HEAPABackbone=Qwen3-0.6B2026.0114.9
GRPO (w. PS)Backbone=Qwen3-0.6B2026.0114.6
GRPOBackbone=Qwen3-0.6B2026.0114.2