Share your thoughts, 1 month free Claude Pro on usSee more

Mathematical Reasoning on DynaMath (Pass@1)

83.3Pass@1

Base

Updated 6d ago

Evaluation Results

Method	Links
Base 2026.05		83.3
SFT + AXPO 2026.05		79
SFT + GRPO 2026.05		78.2
SFT + GRPO 2026.05		77.2
GRPO 2026.05		77.2
SFT 2026.05		76.8
Base 2026.05		76.6
SFT + AXPO 2026.05		76.4
SFT 2026.05		76.3
GRPO 2026.05		76.2
Base 2026.05		75.9
SFT + AXPO 2026.05		66.9
SFT + GRPO 2026.05		66.4
GRPO 2026.05		64.9
SFT 2026.05		64.1
Base 2026.05		60.5
PeRL-VL 2025.12		51.01
RL (Conditional Hard Gate) 2025.12		50.52
VL Cogito 2025.12		50.42
ThinkLite7B 2025.12		49.7
RL (Verifiable Rewards) 2025.12		49.62
PeBR-R1-7B 2025.12		47.48
R1ShareVL7B 2025.12		45.23
Base (Qwen2.5-VL-7B-Instruct) 2025.12		43.01
SelfRewarded-R1-7B 2025.12		41.94