Commonsense Question Answering on CMSQA
72.89AccuracyKite
Evaluation Results
| Method | Links | |
|---|---|---|
| KiteBackbone=Llama 3.2–3B2025.09 | 72.89 | |
| DenseBackbone=Llama 3.2–3B2025.09 | 72.23 | |
| CEILBackbone=Llama 3.2–3B2025.09 | 71.65 | |
| KiteBackbone=Qwen 2.5–1.5B2025.09 | 71.25 | |
| BM25Backbone=Llama 3.2–3B2025.09 | 70.92 | |
| DenseBackbone=Qwen 2.5–1.5B2025.09 | 70.18 | |
| CEILBackbone=Qwen 2.5–1.5B2025.09 | 70.05 | |
| BM25Backbone=Qwen 2.5–1.5B2025.09 | 70.02 | |
| RandomBackbone=Llama 3.2–3B2025.09 | 67.22 | |
| RandomBackbone=Qwen 2.5–1.5B2025.09 | 65.9 | |
| KiteBackbone=GPT-Neo 2.7B2025.09 | 46.6 | |
| DenseBackbone=GPT-Neo 2.7B2025.09 | 43.98 | |
| BM25Backbone=GPT-Neo 2.7B2025.09 | 42.91 | |
| RandomBackbone=GPT-Neo 2.7B2025.09 | 42.25 | |
| CEILBackbone=GPT-Neo 2.7B2025.09 | 37.53 |