Question Answering on HotpotQA (R@k and Adaptive Metrics)
3.497R@1KaLMv2 + Qwen3-235B
Evaluation Results
| Method | Links | ||||
|---|---|---|---|---|---|
| KaLMv2 + Qwen3-235BBackbone=Qwen3-235B, Reranking=False2026.03 | 3.497 | 4.09 | 4.225 | — | |
| KaLMv2 + Llama3.3Backbone=Llama3.3, Reranking=True2026.03 | 3.428 | 4.145 | 4.203 | — | |
| KaLMv2 + Qwen3-235BBackbone=Qwen3-235B, Reranking=True2026.03 | 3.341 | 4.141 | 4.194 | — | |
| KaLMv2 + Llama3.3Backbone=Llama3.3, Reranking=False2026.03 | 3.07 | 3.896 | 4.127 | — | |
| MSABackbone=Qwen3-4B-Instruct2026.03 | — | — | — | 4.061 |