List your company

Instruction Following on Instruction Following (test)

0.9NMSEMENTAT (Detailed Prompt)
0.87561.04031.2051.3697Aug 29, 2025

Evaluation results

MethodLinks
MENTAT (Detailed Prompt)LM=GPT-OSS-20B2025.080.90.43
MENTAT (Basic Prompt)LM=GPT-OSS-20B2025.080.950.42
MENTAT (Basic Prompt)-AvgLM=GPT-OSS-20B2025.081.060.38
GEPALM=GPT-OSS-20B2025.081.060.46
Gradient DescentLM=NeoBERT2025.081.080.36
MENTAT (Detailed Prompt)-AvgLM=GPT-OSS-20B2025.081.090.39
Detailed PromptLM=GPT-OSS-20B2025.081.160.33
Basic PromptLM=GPT-OSS-20B2025.081.180.32
MENTAT (Detailed Prompt) PromptLM=GPT-OSS-20B2025.081.240.36
MENTAT (Basic Prompt) PromptLM=GPT-OSS-20B2025.081.250.35
RL Fine-TuningLM=GPT-OSS-20B2025.081.510.37