Question Answering on GPQA D
63.2AccuracyBF16 (Teacher)
Evaluation Results
| Method | Links | |
|---|---|---|
| BF16 (Teacher)Model=Qwen3-4B-Thinking-2507, Precision=BF162026.06 | 63.2 | |
| CKA-QADModel=Qwen3-4B-Thinking-2507, Precision=NVFP42026.06 | 61.1 | |
| QAD (KL-only)Model=Qwen3-4B-Thinking-2507, Precision=NVFP42026.06 | 59.5 | |
| NVFP4 PTQModel=Qwen3-4B-Thinking-2507, Precision=NVFP42026.06 | 58.8 | |
| IOAModel=Qwen2.5-3B, Teacher LLM=OpenAI o12026.02 | 13.74 | |
| IOAModel=LLaMA3.2-3B, Teacher LLM=OpenAI o12026.02 | 12.15 | |
| MADAModel=Qwen2.5-3B, Teacher LLM=OpenAI o12026.02 | 11.93 | |
| Star-AgentsModel=Qwen2.5-3B, Teacher LLM=OpenAI o12026.02 | 11.67 | |
| CasCoDModel=Qwen2.5-3B, Teacher LLM=OpenAI o12026.02 | 11.19 | |
| CounterDistillModel=Qwen2.5-3B, Teacher LLM=OpenAI o12026.02 | 11.05 | |
| DSSModel=Qwen2.5-3B, Teacher LLM=OpenAI o12026.02 | 10.94 | |
| MADAModel=LLaMA3.2-3B, Teacher LLM=OpenAI o12026.02 | 10.87 | |
| Star-AgentsModel=LLaMA3.2-3B, Teacher LLM=OpenAI o12026.02 | 10.64 | |
| LionModel=Qwen2.5-3B, Teacher LLM=OpenAI o12026.02 | 10.46 | |
| CasCoDModel=LLaMA3.2-3B, Teacher LLM=OpenAI o12026.02 | 10.43 | |
| CounterDistillModel=LLaMA3.2-3B, Teacher LLM=OpenAI o12026.02 | 10.28 | |
| DSSModel=LLaMA3.2-3B, Teacher LLM=OpenAI o12026.02 | 10.08 | |
| LionModel=LLaMA3.2-3B, Teacher LLM=OpenAI o12026.02 | 9.72 | |
| LaMiniModel=Qwen2.5-3B, Teacher LLM=OpenAI o12026.02 | 9.65 | |
| Self-InstructModel=Qwen2.5-3B, Teacher LLM=OpenAI o12026.02 | 9.28 | |
| LaMiniModel=LLaMA3.2-3B, Teacher LLM=OpenAI o12026.02 | 8.91 | |
| Self-InstructModel=LLaMA3.2-3B, Teacher LLM=OpenAI o12026.02 | 8.54 | |
| UndistilledModel=Qwen2.5-3B, Teacher LLM=OpenAI o12026.02 | 7.95 | |
| UndistilledModel=LLaMA3.2-3B, Teacher LLM=OpenAI o12026.02 | 7.14 |