List your company

Multi-turn Code-Switched Question Answering on PingPong answerable QA (id-en) (test)

59.6AccuracyQwen3.5-27B
14.422426.151237.8849.6088Jun 18, 2026

Evaluation results

MethodLinks
Qwen3.5-27BSetting=BASE2026.0659.6
Qwen3.5-27BSetting=CANVAS2026.0658.38
Llama-3.1-8BSetting=CANVAS2026.0643.64
Llama-3.1-8BSetting=BASE2026.0637.37
Aya-Expanse-8BSetting=CANVAS2026.0630.91
Mistral-7BSetting=BASE2026.0629.09
Qwen3.5-27BSetting=CANVAS2026.0628.28
Mistral-7BSetting=CANVAS2026.0628.28
Aya-Expanse-8BSetting=BASE2026.0628.28
Mistral-7BSetting=CANVAS2026.0626.67
Mistral-7BSetting=BASE2026.0625.45
Qwen3.5-27BSetting=BASE2026.0625.25
Llama-3.1-8BSetting=CANVAS2026.0623.64
Llama-3.1-8BSetting=BASE2026.0620.81
Aya-Expanse-8BSetting=CANVAS2026.0618.59
Aya-Expanse-8BSetting=BASE2026.0616.16