List your company

Spoken Mathematics Reasoning on VoxEval College 1.0 (test)

65AccuracyCoT
18.230.3542.554.65Mar 15, 2026

Evaluation results

MethodLinks
CoTModel=Qwen2.5, Hyperparams. (k, α)=-2026.0365
SGSModel=Qwen2.5, Hyperparams. (k, α)=4, 0.0252026.0365
TGSModel=Qwen2.5, Hyperparams. (k, α)=3, 0.052026.0364
VanillaModel=Qwen2.5, Hyperparams. (k, α)=3, 0.0252026.0356
NormalModel=Qwen2.5, Hyperparams. (k, α)=-2026.0340
CoTModel=Phi4-mm, Hyperparams. (k, α)=-2026.0336
VanillaModel=Phi4-mm, Hyperparams. (k, α)=5, 0.0252026.0335
VanillaModel=Voxtral, Hyperparams. (k, α)=5, 0.0252026.0332
SGSModel=Phi4-mm, Hyperparams. (k, α)=3, 0.052026.0330
TGSModel=Phi4-mm, Hyperparams. (k, α)=3, 0.0252026.0330
SGSModel=Voxtral, Hyperparams. (k, α)=3, 0.12026.0329.3
NormalModel=Voxtral, Hyperparams. (k, α)=-2026.0329
CoTModel=AF3, Hyperparams. (k, α)=-2026.0327.4
CoTModel=Voxtral, Hyperparams. (k, α)=-2026.0326.3
TGSModel=Voxtral, Hyperparams. (k, α)=4, 0.0252026.0326.3
NormalModel=AF3, Hyperparams. (k, α)=-2026.0326.1
NormalModel=Phi4-mm, Hyperparams. (k, α)=-2026.0326
SGSModel=AF3, Hyperparams. (k, α)=1, 0.0252026.0324
VanillaModel=AF3, Hyperparams. (k, α)=3, 0.12026.0320
TGSModel=AF3, Hyperparams. (k, α)=5, 0.12026.0320