List your company

Mathematical Reasoning on s1K curated (eval)

48.9AccuracyED-iDPO
15.93224.49133.0541.609May 11, 2026

Evaluation results

MethodLinks
ED-iDPOBackbone=Qwen2.5-7B-Instruct-SFT, Decoding Strategy=Self-Consistency2026.0548.914.3
ED-GRPOBackbone=Qwen2.5-7B-Instruct-SFT, Decoding Strategy=Self-Consistency2026.0548.413.8
DAPOBackbone=Qwen2.5-7B-Instruct-SFT, Decoding Strategy=Self-Consistency2026.0548.113.5
ETOBackbone=Qwen2.5-7B-Instruct-SFT, Decoding Strategy=Self-Consistency2026.0547.312.7
GRPOBackbone=Qwen2.5-7B-Instruct-SFT, Decoding Strategy=Self-Consistency2026.0546.712.1
CoTBackbone=Qwen2.5-7B-Instruct-SFT, Decoding Strategy=Self-Consistency2026.0546.111.5
ED-GRPOBackbone=Qwen2.5-7B-Instruct-SFT, Decoding Strategy=Greedy2026.0538.64
ETOBackbone=Qwen2.5-7B-Instruct-SFT, Decoding Strategy=Greedy2026.05372.4
ED-iDPOBackbone=Qwen2.5-7B-Instruct-SFT, Decoding Strategy=Greedy2026.0536.92.3
DAPOBackbone=Qwen2.5-7B-Instruct-SFT, Decoding Strategy=Greedy2026.0536.11.5
GRPOBackbone=Qwen2.5-7B-Instruct-SFT, Decoding Strategy=Greedy2026.0535.10.5
CoTBackbone=Qwen2.5-7B-Instruct-SFT, Decoding Strategy=Greedy2026.0534.6—
ED-iDPOBackbone=LLaMA3.1-8B-Instruct-SFT, Decoding Strategy=Self-Consistency2026.0528.611.4
ED-GRPOBackbone=LLaMA3.1-8B-Instruct-SFT, Decoding Strategy=Self-Consistency2026.0528.110.9
GRPOBackbone=LLaMA3.1-8B-Instruct-SFT, Decoding Strategy=Self-Consistency2026.0527.310.1
CoTBackbone=LLaMA3.1-8B-Instruct-SFT, Decoding Strategy=Self-Consistency2026.05279.8
ETOBackbone=LLaMA3.1-8B-Instruct-SFT, Decoding Strategy=Self-Consistency2026.0526.79.5
DAPOBackbone=LLaMA3.1-8B-Instruct-SFT, Decoding Strategy=Self-Consistency2026.0526.79.5
ED-GRPOBackbone=LLaMA3.1-8B-Instruct-SFT, Decoding Strategy=Greedy2026.05213.8
GRPOBackbone=LLaMA3.1-8B-Instruct-SFT, Decoding Strategy=Greedy2026.0518.91.7
ED-iDPOBackbone=LLaMA3.1-8B-Instruct-SFT, Decoding Strategy=Greedy2026.0518.71.5
DAPOBackbone=LLaMA3.1-8B-Instruct-SFT, Decoding Strategy=Greedy2026.0518.31.1
ETOBackbone=LLaMA3.1-8B-Instruct-SFT, Decoding Strategy=Greedy2026.0517.70.5
CoTBackbone=LLaMA3.1-8B-Instruct-SFT, Decoding Strategy=Greedy2026.0517.2—