List your company
ResearchFormerly WizWand

Mathematical Reasoning on AMC23 (val)

80Accuracyλ-GRPO
4.624.17543.7563.325Sep 25, 2025Oct 28, 2025Nov 30, 2025Jan 2, 2026Feb 4, 2026Mar 9, 2026Apr 12, 2026
Updated 6mo ago

Evaluation results

MethodLinks
λ-GRPOModel=DeepSeek-R1-Distill-Qwen-1.5B, beta=0.042025.0980
BaseModel=DeepSeek-R1-Distill-Qwen-1.5B, beta=N/A2025.0975
λ-GRPOModel=DeepSeek-R1-Distill-Qwen-1.5B, beta=02025.0975
GRPOModel=DeepSeek-R1-Distill-Qwen-1.5B, beta=0.042025.0975
GenACAlgorithm=GenAC, Base Model=Qwen3-8B-Base, Sampled generations=16, RL Setup=Zero RL2026.0474.53
VC-PPOAlgorithm=VC-PPO, Base Model=Qwen3-8B-Base, Sampled generations=16, RL Setup=Zero RL2026.0470.78
GRPOAlgorithm=GRPO, Base Model=Qwen3-8B-Base, Sampled generations=16, RL Setup=Zero RL2026.0469.69
RLOOAlgorithm=RLOO, Base Model=Qwen3-8B-Base, Sampled generations=16, RL Setup=Zero RL2026.0468.75
GRPOModel=DeepSeek-R1-Distill-Qwen-1.5B, beta=02025.0962.5
OBLR-POModel=Qwen3-4B-Base2025.1155
RLOOModel=Qwen3-8B-Base2025.1155
PPOModel=Qwen3-8B-Base2025.1152.5
ReMaxModel=Qwen3-4B-Base2025.1150
GRPOModel=Qwen3-4B-Base2025.1147.5
PPOModel=Qwen3-4B-Base2025.1147.5
ReMaxModel=Qwen3-8B-Base2025.1147.5
OBLR-POModel=Qwen3-8B-Base2025.1147.5
RLOOModel=Qwen3-4B-Base2025.1142.5
GRPOModel=Qwen3-8B-Base2025.1142.5
GRPOModel=Llama-3.2-1B-Instruct, beta=0.042025.0917.5
λ-GRPOModel=Llama-3.2-1B-Instruct, beta=02025.0912.5
BaseModel=Llama-3.2-1B-Instruct, beta=N/A2025.097.5
GRPOModel=Llama-3.2-1B-Instruct, beta=02025.097.5
λ-GRPOModel=Llama-3.2-1B-Instruct, beta=0.042025.097.5