Machine Translation on WMT EN-TR (test)
11.7BLEUMix-MoE
Evaluation results
| Method | Links | |||
|---|---|---|---|---|
| Mix-MoEBase Model=Llama3.2-1B, Training Strategy=Mix-MoE2026.05 | 11.7 | 23.1 | 66.8 | |
| MOE-LPRBase Model=Llama3.2-1B, Training Strategy=MOE-LPR2026.05 | 9.9 | 20.4 | 65 | |
| MoEBase Model=Llama3.2-1B, Training Strategy=MoE2026.05 | 8.8 | 17.5 | 65.3 | |
| Qwen2.5-1.5BBase Model=Qwen2.5-1.5B2026.05 | 7.6 | 14.7 | 63.4 | |
| Dense PT+SFTBase Model=Llama3.2-1B, Training Strategy=Dense PT+SFT2026.05 | 7.3 | 15 | 57 | |
| Full FinetuneBase Model=Llama3.2-1B, Training Strategy=Full Finetune2026.05 | 6.3 | 13.1 | 53.9 | |
| Qwen2.5-0.5BBase Model=Qwen2.5-0.5B2026.05 | 5.1 | 8.5 | 50.5 | |
| LLAMA-ProBase Model=Llama3.2-1B, Training Strategy=LLAMA-Pro2026.05 | 4.1 | 9.2 | 47.8 | |
| LLAMA-MOE v2Base Model=Llama3.2-1B, Training Strategy=LLAMA-MOE v22026.05 | 3.9 | 8.7 | 51.5 | |
| Llama3.2-1BBase Model=Llama3.2-1B2026.05 | 3.6 | 6 | 47.5 | |
| LoRABase Model=Llama3.2-1B, Training Strategy=LoRA2026.05 | 3 | 4.7 | 45.3 |