List your company

Instruction Following on Alpaca SFT Small Prompts

46.6Self-BLEUCE
1.04812.87424.736.526Apr 30, 2026

Evaluation results

MethodLinks
CEBackbone=PHI-4-14B2026.0446.64.2
CEBackbone=OLMo-2-13B2026.04454.3
CEBackbone=QWEN-3-8B2026.0444.94.1
CEBackbone=Mistral-12B2026.0444.54.2
CEBackbone=LLAMA-3.1-8B2026.0444.54.2
CEBackbone=Pythia-12B2026.0435.93.8
BASEBackbone=QWEN-3-8B2026.0431.53.8
GEMBackbone=QWEN-3-8B2026.0429.54
FLBackbone=Mistral-12B2026.0429.34.1
FLBackbone=QWEN-3-8B2026.0429.24
FLBackbone=OLMo-2-13B2026.04294.2
FLBackbone=PHI-4-14B2026.0428.94.2
FLBackbone=LLAMA-3.1-8B2026.0428.34.1
GEMBackbone=Mistral-12B2026.0427.74.1
GEMBackbone=PHI-4-14B2026.0427.54.1
GEMBackbone=OLMo-2-13B2026.04274.1
GEMBackbone=LLAMA-3.1-8B2026.0425.94.1
TOFUBackbone=QWEN-3-8B2026.0424.14
FLBackbone=Pythia-12B2026.0422.23.6
TOFUBackbone=PHI-4-14B2026.0421.84.1
TOFUBackbone=OLMo-2-13B2026.0421.34.1
TOFUBackbone=Mistral-12B2026.0421.24
TOFUBackbone=LLAMA-3.1-8B2026.0420.94
GEMBackbone=Pythia-12B2026.0417.63.5
BASEBackbone=PHI-4-14B2026.0417.63.9
TOFUBackbone=Pythia-12B2026.0416.13.5
BASEBackbone=OLMo-2-13B2026.0413.93.8
BASEBackbone=Mistral-12B2026.0412.73.8
BASEBackbone=LLAMA-3.1-8B2026.0412.23.5
BASEBackbone=Pythia-12B2026.048.42.6
λ-PRBackbone=QWEN-3-8B2026.044.62.6
λ-PRBackbone=PHI-4-14B2026.043.32.4
λ-PRBackbone=OLMo-2-13B2026.043.22.3
λ-PRBackbone=Pythia-12B2026.0432
λ-PRBackbone=Mistral-12B2026.042.92
λ-PRBackbone=LLAMA-3.1-8B2026.042.82.1