Machine Translation on English-to-Wolof
22.53chrFOur RL
Evaluation results
| Method | Links | |
|---|---|---|
| Our RLBase Model=Qwen3-4B-Base, Training Method=RL, Context=full2026.06 | 22.53 | |
| Our RLBase Model=Llama-3.2-3B-Instruct, Training Method=RL, Context=full2026.06 | 20.29 | |
| Qwen3-4B-BaseBase Model=Qwen3-4B-Base, Training Method=Base (untuned), Context=full2026.06 | 15.05 | |
| Qwen3-4B-BaseBase Model=Qwen3-4B-Base, Training Method=Base (untuned), Context=none2026.06 | 13.03 | |
| Our RLBase Model=Llama-3.2-3B-Instruct, Training Method=RL, Context=none2026.06 | 11.34 | |
| Our SFTBase Model=Llama-3.2-3B-Instruct, Training Method=SFT, Context=none2026.06 | 11.31 | |
| Llama-3.2-3B-InstBase Model=Llama-3.2-3B-Instruct, Training Method=Instruct-tuned, Context=full2026.06 | 11.18 | |
| Our RLBase Model=Qwen3-4B-Base, Training Method=RL, Context=none2026.06 | 10.25 | |
| Our SFTBase Model=Llama-3.2-3B-Instruct, Training Method=SFT, Context=full2026.06 | 5.76 | |
| Llama-3.2-3B-InstBase Model=Llama-3.2-3B-Instruct, Training Method=Instruct-tuned, Context=none2026.06 | 5.71 | |
| Our SFTBase Model=Qwen3-4B-Base, Training Method=SFT, Context=full2026.06 | 4.84 | |
| Our SFTBase Model=Qwen3-4B-Base, Training Method=SFT, Context=none2026.06 | 4.72 |