Knowledge Tracing on EdNet-500 (test)
0.81AUCAKT
Evaluation results
| Method | Links | |||
|---|---|---|---|---|
| AKTtraining=traditional, design=–2026.01 | 0.81 | — | — | |
| DKTtraining=traditional, design=–2026.01 | 0.789 | — | — | |
| DKVMNtraining=traditional, design=–2026.01 | 0.78 | — | — | |
| Gemini-2.5-flashdesign=No-Option, framework=Thinking-KT2026.01 | 0.6969 | 0.7388 | 0.8219 | |
| GPT-5-minidesign=Weight, framework=Thinking-KT2026.01 | 0.6327 | 0.7692 | 0.8536 | |
| GPT-5-minidesign=Option, framework=LLM-based (No-Think)2026.01 | 0.6063 | 0.6132 | 0.6892 | |
| Gemini-2.5-flashdesign=Weight, framework=LLM-based (No-Think)2026.01 | 0.6058 | 0.5511 | 0.5465 | |
| GPT-5-minidesign=Option, framework=Thinking-KT2026.01 | 0.6053 | 0.6124 | 0.6871 | |
| Qwen3-1.7Bdesign=No-Option, framework=Thinking-KT2026.01 | 0.6046 | 0.5792 | 0.6209 | |
| Qwen3-1.7Bdesign=Option, framework=Thinking-KT2026.01 | 0.6043 | 0.5932 | 0.6622 | |
| Gemini-2.5-flashdesign=Weight, framework=Thinking-KT2026.01 | 0.6004 | 0.5551 | 0.5524 | |
| GPT-5-minidesign=No-Option, framework=Thinking-KT2026.01 | 0.5992 | 0.5935 | 0.6666 | |
| GPT-5-minidesign=Weight, framework=LLM-based (No-Think)2026.01 | 0.5889 | 0.6016 | 0.671 | |
| Gemini-2.5-flashdesign=Option, framework=LLM-based (No-Think)2026.01 | 0.5884 | 0.5731 | 0.6048 | |
| Gemini-2.5-flashdesign=Option, framework=Thinking-KT2026.01 | 0.5879 | 0.5671 | 0.6 | |
| Gemini-2.5-flashdesign=No-Option, framework=LLM-based (No-Think)2026.01 | 0.5875 | 0.5671 | 0.5781 | |
| Qwen3-1.7Bdesign=Weight, framework=Thinking-KT2026.01 | 0.5612 | 0.5531 | 0.6135 | |
| GPT-5-minidesign=No-Option, framework=LLM-based (No-Think)2026.01 | 0.5117 | 0.4068 | 0.0142 | |
| Qwen3-1.7Bdesign=Option, framework=LLM-based (No-Think)2026.01 | 0.5096 | 0.5783 | 0.7075 | |
| Qwen3-1.7Bdesign=Weight, framework=LLM-based (No-Think)2026.01 | 0.5042 | 0.6012 | 0.7497 | |
| Qwen3-1.7Bdesign=No-Option, framework=LLM-based (No-Think)2026.01 | 0.493 | 0.5511 | 0.6716 |