List your company

Instruction Following on AlpacaEval LC (test)

84.8Win RateDeepSeek-V3
27.70442.52757.3572.173Feb 18, 2026

Evaluation results

MethodLinks
DeepSeek-V3Backbone=DeepSeek-V32026.0284.8
RefEvalBackbone=Llama3-8B-Inst2026.0273.1
RefEvalBackbone=Qwen2.5-7B2026.0270
SimPOBackbone=Llama3-8B-Inst2026.0251.6
Qwen2.5-7B-InstBackbone=Qwen2.5-7B2026.0229.9