List your company

Mathematics on GSM8K 0 (test)

83.7AccuracyESPO
70.80474.15277.580.848Dec 3, 2025

Evaluation results

MethodLinks
ESPOBackbone=LLaDA-8B, Seq Len=512, Zero-shot=true2025.1283.7
wd1Backbone=Dream-7B, Seq Len=512, Zero-shot=true2025.1282.9
wd1Backbone=Dream-7B, Seq Len=256, Zero-shot=true2025.1282.4
wd1Backbone=LLaDA-8B, Seq Len=512, Zero-shot=true2025.1282.3
ESPOBackbone=LLaDA-8B, Seq Len=256, Zero-shot=true2025.1282.3
ESPOBackbone=Dream-7B, Seq Len=256, Zero-shot=true2025.1282.3
ESPOBackbone=LLaDA-8B, Seq Len=Avg., Zero-shot=true2025.1282
ESPOBackbone=Dream-7B, Seq Len=512, Zero-shot=true2025.1282
diffu-GRPOBackbone=Dream-7B, Seq Len=256, Zero-shot=true2025.1281.9
diffu-GRPOBackbone=Dream-7B, Seq Len=512, Zero-shot=true2025.1281.7
DreamBackbone=Dream-7B, Seq Len=256, Zero-shot=true2025.1281.3
ESPOBackbone=Dream-7B, Seq Len=Avg., Zero-shot=true2025.1281.3
diffu-GRPOBackbone=LLaDA-8B, Seq Len=512, Zero-shot=true2025.1281.2
wd1Backbone=LLaDA-8B, Seq Len=256, Zero-shot=true2025.1280.8
DreamBackbone=Dream-7B, Seq Len=512, Zero-shot=true2025.1280.7
wd1Backbone=Dream-7B, Seq Len=Avg., Zero-shot=true2025.1280.5
LLaDABackbone=LLaDA-8B, Seq Len=512, Zero-shot=true2025.1280.2
diffu-GRPOBackbone=Dream-7B, Seq Len=Avg., Zero-shot=true2025.1280.2
wd1Backbone=LLaDA-8B, Seq Len=Avg., Zero-shot=true2025.1280.1
ESPOBackbone=LLaDA-8B, Seq Len=128, Zero-shot=true2025.1280
ESPOBackbone=Dream-7B, Seq Len=128, Zero-shot=true2025.1279.6
DreamBackbone=Dream-7B, Seq Len=Avg., Zero-shot=true2025.1279.3
diffu-GRPOBackbone=LLaDA-8B, Seq Len=256, Zero-shot=true2025.1278.1
diffu-GRPOBackbone=LLaDA-8B, Seq Len=Avg., Zero-shot=true2025.1278
wd1Backbone=LLaDA-8B, Seq Len=128, Zero-shot=true2025.1277.2
diffu-GRPOBackbone=Dream-7B, Seq Len=128, Zero-shot=true2025.1277
wd1Backbone=Dream-7B, Seq Len=128, Zero-shot=true2025.1276.3
LLaDABackbone=LLaDA-8B, Seq Len=256, Zero-shot=true2025.1276.2
LLaDABackbone=LLaDA-8B, Seq Len=Avg., Zero-shot=true2025.1275.9
DreamBackbone=Dream-7B, Seq Len=128, Zero-shot=true2025.1275.8
diffu-GRPOBackbone=LLaDA-8B, Seq Len=128, Zero-shot=true2025.1274.6
LLaDABackbone=LLaDA-8B, Seq Len=128, Zero-shot=true2025.1271.3