Mathematical Reasoning on MGSM Code Switched P, I - (EN), Q(X) (Avg)
100Language ConsistencyBloom-7.1B (Full scope SFT)
Evaluation results
| Method | Links | ||
|---|---|---|---|
| Bloom-7.1B (Full scope SFT)Model=Bloom-7.1B, Fine-tuning Strategy=Full scope SFT2026.01 | 100 | 2.2 | |
| Bloom-7.1B (Selective SFT)Model=Bloom-7.1B, Fine-tuning Strategy=Selective SFT2026.01 | 99.6 | 2 | |
| Qwen-3-32B (Selective SFT)Model=Qwen-3-32B, Fine-tuning Strategy=Selective SFT2026.01 | 98.6 | 84.6 | |
| Qwen-3-32B (Full scope SFT)Model=Qwen-3-32B, Fine-tuning Strategy=Full scope SFT2026.01 | 95 | 87 | |
| Bloom-7.1B (Random Selective SFT)Model=Bloom-7.1B, Fine-tuning Strategy=Random Selective SFT2026.01 | 68 | 0 | |
| Qwen-3-32B (Random Selective SFT)Model=Qwen-3-32B, Fine-tuning Strategy=Random Selective SFT2026.01 | 53.8 | 0 | |
| Bloom-7.1B (Pre-Finetuning)Model=Bloom-7.1B, Fine-tuning Strategy=Pre-Finetuning2026.01 | 18.4 | 0.4 | |
| Qwen-3-32B (Pre-Finetuning)Model=Qwen-3-32B, Fine-tuning Strategy=Pre-Finetuning2026.01 | 6.8 | 57 |