General Capability on BBH, GSM8K, MMLU, TruthfulQA, HumanEval, MBPP
26.77Average ScoreADG
Evaluation Results
| Method | Links | |
|---|---|---|
| ADGBackbone=LLaMA3.2-1B-Instruct, Selection Pool=Alpaca-GPT4(52K), Selection Budget=10K2026.04 | 26.77 | |
| ADGBackbone=LLaMA3.2-1B-Instruct, Selection Pool=CoT(100K), Selection Budget=10K2026.04 | 26.23 | |
| ADGBackbone=LLaMA3.2-1B-Instruct, Selection Pool=WizardLM(70K), Selection Budget=10K2026.04 | 26.13 | |
| SuperFilteringBackbone=LLaMA3.2-1B-Instruct, Selection Pool=Alpaca-GPT4(52K), Selection Budget=10K2026.04 | 26.03 | |
| ZIPBackbone=LLaMA3.2-1B-Instruct, Selection Pool=Alpaca-GPT4(52K), Selection Budget=10K2026.04 | 25.95 | |
| SelectITBackbone=LLaMA3.2-1B-Instruct, Selection Pool=Alpaca-GPT4(52K), Selection Budget=10K2026.04 | 25.85 | |
| ZIPBackbone=LLaMA3.2-1B-Instruct, Selection Pool=CoT(100K), Selection Budget=10K2026.04 | 25.61 | |
| RethinkingBackbone=LLaMA3.2-1B-Instruct, Selection Pool=Alpaca-GPT4(52K), Selection Budget=10K2026.04 | 25.57 | |
| RethinkingBackbone=LLaMA3.2-1B-Instruct, Selection Pool=CoT(100K), Selection Budget=10K2026.04 | 25.45 | |
| IFDBackbone=LLaMA3.2-1B-Instruct, Selection Pool=Alpaca-GPT4(52K), Selection Budget=10K2026.04 | 25.33 | |
| SelectITBackbone=LLaMA3.2-1B-Instruct, Selection Pool=WizardLM(70K), Selection Budget=10K2026.04 | 25.32 | |
| SCARBackbone=LLaMA3.2-1B-Instruct, Selection Pool=WizardLM(70K), Selection Budget=10K2026.04 | 25.14 | |
| MIGBackbone=LLaMA3.2-1B-Instruct, Selection Pool=WizardLM(70K), Selection Budget=10K2026.04 | 25.06 | |
| RandomBackbone=LLaMA3.2-1B-Instruct, Selection Pool=WizardLM(70K), Selection Budget=10K2026.04 | 25.01 | |
| SCARBackbone=LLaMA3.2-1B-Instruct, Selection Pool=Alpaca-GPT4(52K), Selection Budget=10K2026.04 | 24.99 | |
| SuperFilteringBackbone=LLaMA3.2-1B-Instruct, Selection Pool=CoT(100K), Selection Budget=10K2026.04 | 24.99 | |
| RandomBackbone=LLaMA3.2-1B-Instruct, Selection Pool=CoT(100K), Selection Budget=10K2026.04 | 24.95 | |
| RethinkingBackbone=LLaMA3.2-1B-Instruct, Selection Pool=WizardLM(70K), Selection Budget=10K2026.04 | 24.8 | |
| ZIPBackbone=LLaMA3.2-1B-Instruct, Selection Pool=WizardLM(70K), Selection Budget=10K2026.04 | 24.79 | |
| IFDBackbone=LLaMA3.2-1B-Instruct, Selection Pool=CoT(100K), Selection Budget=10K2026.04 | 24.74 | |
| SCARBackbone=LLaMA3.2-1B-Instruct, Selection Pool=CoT(100K), Selection Budget=10K2026.04 | 24.74 | |
| All DataBackbone=LLaMA3.2-1B-Instruct, Selection Pool=Alpaca-GPT4(52K), Selection Budget=10K2026.04 | 24.73 | |
| SuperFilteringBackbone=LLaMA3.2-1B-Instruct, Selection Pool=WizardLM(70K), Selection Budget=10K2026.04 | 24.73 | |
| SelectITBackbone=LLaMA3.2-1B-Instruct, Selection Pool=CoT(100K), Selection Budget=10K2026.04 | 24.68 | |
| MIGBackbone=LLaMA3.2-1B-Instruct, Selection Pool=Alpaca-GPT4(52K), Selection Budget=10K2026.04 | 24.67 | |
| RandomBackbone=LLaMA3.2-1B-Instruct, Selection Pool=Alpaca-GPT4(52K), Selection Budget=10K2026.04 | 24.55 | |
| IFDBackbone=LLaMA3.2-1B-Instruct, Selection Pool=WizardLM(70K), Selection Budget=10K2026.04 | 24.52 | |
| All DataBackbone=LLaMA3.2-1B-Instruct, Selection Pool=WizardLM(70K), Selection Budget=10K2026.04 | 24.21 | |
| MIGBackbone=LLaMA3.2-1B-Instruct, Selection Pool=CoT(100K), Selection Budget=10K2026.04 | 24.2 | |
| All DataBackbone=LLaMA3.2-1B-Instruct, Selection Pool=CoT(100K), Selection Budget=10K2026.04 | 23.12 |