List your company
ResearchFormerly WizWand

Machine Translation on English-to-Kalamang

34.64chrF ScoreOur RL
11.354417.399723.44529.4903Jun 4, 2026
Updated 5mo ago

Evaluation results

MethodLinks
Our RLBase Model=Qwen3-4B-Base, Training Method=RL, Context=full2026.0634.64
Our RLBase Model=Llama-3.2-3B-Instruct, Training Method=RL, Context=full2026.0630.05
Our SFTBase Model=Qwen3-4B-Base, Training Method=SFT, Context=full2026.0628.6
Qwen3-4B-BaseBase Model=Qwen3-4B-Base, Training Method=Base (untuned), Context=full2026.0625.58
Our SFTBase Model=Llama-3.2-3B-Instruct, Training Method=SFT, Context=full2026.0624.44
Llama-3.2-3B-InstBase Model=Llama-3.2-3B-Instruct, Training Method=Instruct-tuned, Context=full2026.0620.14
Our RLBase Model=Qwen3-4B-Base, Training Method=RL, Context=none2026.0616.87
Our RLBase Model=Llama-3.2-3B-Instruct, Training Method=RL, Context=none2026.0616.69
Llama-3.2-3B-InstBase Model=Llama-3.2-3B-Instruct, Training Method=Instruct-tuned, Context=none2026.0614.85
Qwen3-4B-BaseBase Model=Qwen3-4B-Base, Training Method=Base (untuned), Context=none2026.0614.7
Our SFTBase Model=Qwen3-4B-Base, Training Method=SFT, Context=none2026.0613.58
Our SFTBase Model=Llama-3.2-3B-Instruct, Training Method=SFT, Context=none2026.0612.25