Common Sense Reasoning on WinoGrande (ET/ES/E*S/AR% Evaluation)
64.5E*SRecover-LoRA
Evaluation Results
| Method | Links | ||||
|---|---|---|---|---|---|
| Recover-LoRAModel=Qwen3-4B, Quantization=W2-GateUp, Training Data Source=Synthetic-10k, Training Samples=10k2026.06 | 64.5 | — | — | 88.4 | |
| Recover-LoRAModel=Qwen3-4B, Quantization=W2-GateUp, Training Data Source=OpenHermes-10k, Training Samples=10k2026.06 | 64 | — | — | 85.3 | |
| Quantized Student (W2-GateUp)Model=Qwen3-4B, Quantization=W2-GateUp (gate/up at INT2)2026.06 | — | — | 50.7 | — | |
| Teacher (BF16)Model=Qwen3-4B, Precision=BF162026.06 | — | 66.3 | — | — |