Multi-task Language Understanding on TinyMMLU
66.1E*S Combined ScoreRecover-LoRA
Evaluation results
| Method | Links | ||||
|---|---|---|---|---|---|
| Recover-LoRAModel=Qwen3-4B, Quantization=W2-GateUp, Training Data Source=OpenHermes-10k, Training Samples=10k2026.06 | 66.1 | — | — | 92.7 | |
| Recover-LoRAModel=Qwen3-4B, Quantization=W2-GateUp, Training Data Source=Synthetic-10k, Training Samples=10k2026.06 | 60.5 | — | — | 79.4 | |
| Quantized Student (W2-GateUp)Model=Qwen3-4B, Quantization=W2-GateUp (gate/up at INT2)2026.06 | — | — | 26.9 | — | |
| Teacher (BF16)Model=Qwen3-4B, Precision=BF162026.06 | — | 69.2 | — | — |