List your company
ResearchFormerly WizWand

Language Modeling on OpenOrca (upper-level)

0.76Evaluation LossOnline (R = 10%, G = 5)
0.7280.9441.161.376Nov 26, 2025
Updated 4mo ago

Evaluation results

MethodLinks
Online (R = 10%, G = 5)Base LLM=LLAMA-8B, Sample ratio (R)=10%, Generations (G)=52025.110.76
Online (R = 10%, G = 1)Base LLM=LLAMA-8B, Sample ratio (R)=10%, Generations (G)=12025.110.78
Offline BDSBase LLM=LLAMA-8B2025.110.8
Online (R = 5%, G = 1)Base LLM=LLAMA-8B, Sample ratio (R)=5%, Generations (G)=12025.110.8
Direct mixing (ρ = 0.5)Base LLM=LLAMA-8B2025.110.84
LESSBase LLM=LLAMA-8B2025.110.88
Direct mixing (ρ = 1)Base LLM=LLAMA-8B2025.110.92
DSIRBase LLM=LLAMA-8B2025.110.92
Random selectionBase LLM=LLAMA-8B2025.110.94
Online (R = 10%, G = 5)Base LLM=PYTHIA-1B, Sample ratio (R)=10%, Generations (G)=52025.111.32
Online (R = 10%, G = 1)Base LLM=PYTHIA-1B, Sample ratio (R)=10%, Generations (G)=12025.111.34
Online (R = 5%, G = 1)Base LLM=PYTHIA-1B, Sample ratio (R)=5%, Generations (G)=12025.111.37
Offline BDSBase LLM=PYTHIA-1B2025.111.38
Direct mixing (ρ = 0.5)Base LLM=PYTHIA-1B2025.111.41
DSIRBase LLM=PYTHIA-1B2025.111.52
LESSBase LLM=PYTHIA-1B2025.111.54
Random selectionBase LLM=PYTHIA-1B2025.111.55
Direct mixing (ρ = 1)Base LLM=PYTHIA-1B2025.111.56