List your company

Next Token Prediction on OpenWebText (held-out)

18.53ID PPLConcepts
-877.7025,171.86411,221.4317,270.996Mar 31, 2026

Evaluation results

MethodLinks
ConceptsModel=Llama 3.1 8B, λ=0.752026.0318.5340.13
ConceptsModel=Llama 3.1 8B, λ=0.52026.0318.8939.45
ConceptsModel=Llama 3.1 8B, λ=1.02026.0318.9239.69
ConceptsModel=Llama 3.1 8B, λ=0.252026.0319.8438.31
BaselineModel=Llama 3.1 8B, λ=0.02026.0322.0536.44
BaselineModel=Llama 3.1 8B, λ=Base2026.0322.2636.4
ConceptsModel=Llama 3.2 3B, λ=1.02026.0322.7737.64
ConceptsModel=Llama 3.2 3B, λ=0.752026.0322.8737.41
ConceptsModel=Llama 3.2 3B, λ=0.52026.0323.6336.58
ConceptsModel=Llama 3.2 3B, λ=0.252026.0325.0935.35
BaselineModel=Llama 3.2 3B, λ=0.02026.0328.0333.43
BaselineModel=Llama 3.2 3B, λ=Base2026.0328.4333.47
ConceptsModel=Llama 3.2 1B, λ=1.02026.0330.334.28
ConceptsModel=Llama 3.2 1B, λ=0.752026.0330.6533.88
ConceptsModel=Llama 3.2 1B, λ=0.52026.0331.8533.13
ConceptsModel=Llama 3.2 1B, λ=0.252026.0333.9432
BaselineModel=Llama 3.2 1B, λ=0.02026.0338.0530.24
BaselineModel=Llama 3.2 1B, λ=Base2026.0338.5630.11
NoiseModel=Llama 3.1 8B, λ=0.252026.0339.2233.62
NoiseModel=Llama 3.2 3B, λ=0.252026.0348.6930.47
NoiseModel=Llama 3.2 1B, λ=0.252026.0363.9726.75
NoiseModel=Llama 3.1 8B, λ=0.52026.0377.7330.67
NoiseModel=Llama 3.2 3B, λ=0.52026.03100.4626.63
NoiseModel=Llama 3.2 1B, λ=0.52026.03132.8422.46
NoiseModel=Llama 3.1 8B, λ=0.752026.03202.2124.45
NoiseModel=Llama 3.2 3B, λ=0.752026.03275.9619.33
NoiseModel=Llama 3.2 1B, λ=0.752026.03377.8915.58
NoiseModel=Llama 3.2 1B, λ=1.02026.0318,020.530.47
NoiseModel=Llama 3.2 3B, λ=1.02026.0318,376.560.68
NoiseModel=Llama 3.1 8B, λ=1.02026.0322,424.330.49