Language Understanding on MMLU (Code Switched P, I - (EN), Q(X))
99.87Language ConsistencyQwen-3-32B (Full scope SFT)
Evaluation results
| Method | Links | ||
|---|---|---|---|
| Qwen-3-32B (Full scope SFT)Model=Qwen-3-32B, Fine-tuning Strategy=Full scope SFT2026.01 | 99.87 | 78.84 | |
| Bloom-7.1B (Full scope SFT)Model=Bloom-7.1B, Fine-tuning Strategy=Full scope SFT2026.01 | 99.87 | 33.72 | |
| Qwen-3-32B (Selective SFT)Model=Qwen-3-32B, Fine-tuning Strategy=Selective SFT2026.01 | 99.62 | 74.44 | |
| Bloom-7.1B (Selective SFT)Model=Bloom-7.1B, Fine-tuning Strategy=Selective SFT2026.01 | 98.66 | 21.14 | |
| Qwen-3-32B (Random Selective SFT)Model=Qwen-3-32B, Fine-tuning Strategy=Random Selective SFT2026.01 | 98.3 | 1.67 | |
| Bloom-7.1B (Random Selective SFT)Model=Bloom-7.1B, Fine-tuning Strategy=Random Selective SFT2026.01 | 55.56 | 0 | |
| Bloom-7.1B (Pre-Finetuning)Model=Bloom-7.1B, Fine-tuning Strategy=Pre-Finetuning2026.01 | 29.49 | 22.31 | |
| Qwen-3-32B (Pre-Finetuning)Model=Qwen-3-32B, Fine-tuning Strategy=Pre-Finetuning2026.01 | 8.35 | 60.51 |