Question Answering on TQA (test)
90.2AUROCSemantic Token Clustering
Evaluation Results
| Method | Links | |
|---|---|---|
| Semantic Token ClusteringLLM Backbone=Qwen2.5-7B2026.03 | 90.2 | |
| Semantic Token ClusteringLLM Backbone=Qwen2.5-14B2026.03 | 90 | |
| CCPLLM Backbone=Qwen2.5-7B2026.03 | 89.5 | |
| Semantic Token ClusteringLLM Backbone=Mistral-7B2026.03 | 89.3 | |
| CCPLLM Backbone=Mistral-7B2026.03 | 89.2 | |
| PerplexityLLM Backbone=Qwen2.5-7B2026.03 | 88.8 | |
| CCPLLM Backbone=Qwen2.5-14B2026.03 | 88.8 | |
| SARLLM Backbone=Qwen2.5-7B2026.03 | 88.4 | |
| LN EntropyLLM Backbone=Qwen2.5-7B2026.03 | 88.3 | |
| sentenceSARLLM Backbone=Qwen2.5-7B2026.03 | 88.3 | |
| Semantic Token ClusteringLLM Backbone=Llama-3-8B2026.03 | 88.2 | |
| PerplexityLLM Backbone=Qwen2.5-14B2026.03 | 88.1 | |
| SARLLM Backbone=Mistral-7B2026.03 | 88.1 | |
| Predictive EntropyLLM Backbone=Qwen2.5-7B2026.03 | 88 | |
| SARLLM Backbone=Qwen2.5-14B2026.03 | 87.9 | |
| CCPLLM Backbone=Llama-3-8B2026.03 | 87.8 | |
| Semantic Token ClusteringLLM Backbone=Llama-2-7B2026.03 | 87.8 | |
| sentenceSARLLM Backbone=Qwen2.5-14B2026.03 | 87.7 | |
| Semantic Token ClusteringLLM Backbone=Qwen2.5-3B2026.03 | 87.6 | |
| LN EntropyLLM Backbone=Qwen2.5-14B2026.03 | 87.5 | |
| CCPLLM Backbone=Llama-2-7B2026.03 | 87.4 | |
| Predictive EntropyLLM Backbone=Qwen2.5-14B2026.03 | 87.3 | |
| PerplexityLLM Backbone=Mistral-7B2026.03 | 87.2 | |
| SARLLM Backbone=Llama-3-8B2026.03 | 87.2 | |
| tokenSARLLM Backbone=Qwen2.5-7B2026.03 | 87.1 | |
| LN EntropyLLM Backbone=Mistral-7B2026.03 | 87 | |
| P(True)LLM Backbone=Qwen2.5-14B2026.03 | 86.9 | |
| sentenceSARLLM Backbone=Mistral-7B2026.03 | 86.9 | |
| CCPLLM Backbone=Qwen2.5-3B2026.03 | 86.9 | |
| PerplexityLLM Backbone=Llama-2-7B2026.03 | 86.8 | |
| tokenSARLLM Backbone=Qwen2.5-14B2026.03 | 86.7 | |
| SARLLM Backbone=Llama-2-7B2026.03 | 86.7 | |
| PerplexityLLM Backbone=Llama-3-8B2026.03 | 86.5 | |
| Predictive EntropyLLM Backbone=Mistral-7B2026.03 | 86.4 | |
| LN EntropyLLM Backbone=Llama-2-7B2026.03 | 86.2 | |
| LN EntropyLLM Backbone=Llama-3-8B2026.03 | 86.2 | |
| tokenSARLLM Backbone=Mistral-7B2026.03 | 86 | |
| sentenceSARLLM Backbone=Llama-2-7B2026.03 | 85.9 | |
| sentenceSARLLM Backbone=Llama-3-8B2026.03 | 85.9 | |
| PerplexityLLM Backbone=Qwen2.5-3B2026.03 | 85.6 | |
| Predictive EntropyLLM Backbone=Llama-2-7B2026.03 | 85.5 | |
| Predictive EntropyLLM Backbone=Llama-3-8B2026.03 | 85.4 | |
| sentenceSARLLM Backbone=Qwen2.5-3B2026.03 | 85.4 | |
| SARLLM Backbone=Qwen2.5-3B2026.03 | 85.4 | |
| LN EntropyLLM Backbone=Qwen2.5-3B2026.03 | 85.3 | |
| tokenSARLLM Backbone=Llama-2-7B2026.03 | 85.3 | |
| tokenSARLLM Backbone=Llama-3-8B2026.03 | 85.3 | |
| Predictive EntropyLLM Backbone=Qwen2.5-3B2026.03 | 85 | |
| EigVLLM Backbone=Qwen2.5-7B2026.03 | 84.9 | |
| EigVLLM Backbone=Mistral-7B2026.03 | 84.5 | |
| Semantic EntropyLLM Backbone=Qwen2.5-7B2026.03 | 84.4 | |
| Semantic EntropyLLM Backbone=Mistral-7B2026.03 | 83.9 | |
| DegLLM Backbone=Qwen2.5-7B2026.03 | 83.9 | |
| EigVLLM Backbone=Qwen2.5-3B2026.03 | 83.7 | |
| DegLLM Backbone=Mistral-7B2026.03 | 83.7 | |
| EigVLLM Backbone=Qwen2.5-14B2026.03 | 83.5 | |
| P(True)LLM Backbone=Qwen2.5-7B2026.03 | 83.2 | |
| Semantic EntropyLLM Backbone=Qwen2.5-3B2026.03 | 83.2 | |
| Semantic EntropyLLM Backbone=Qwen2.5-14B2026.03 | 83.2 | |
| EigVLLM Backbone=Llama-2-7B2026.03 | 83.2 | |
| Eigen ScoreLLM Backbone=Qwen2.5-7B2026.03 | 83.2 | |
| tokenSARLLM Backbone=Qwen2.5-3B2026.03 | 83.1 | |
| DegLLM Backbone=Qwen2.5-3B2026.03 | 83.1 | |
| DegLLM Backbone=Qwen2.5-14B2026.03 | 82.7 | |
| Semantic EntropyLLM Backbone=Llama-2-7B2026.03 | 82.6 | |
| EigVLLM Backbone=Llama-3-8B2026.03 | 82.5 | |
| DegLLM Backbone=Llama-2-7B2026.03 | 82.4 | |
| Semantic EntropyLLM Backbone=Llama-3-8B2026.03 | 82.3 | |
| Eigen ScoreLLM Backbone=Qwen2.5-3B2026.03 | 82 | |
| DegLLM Backbone=Llama-3-8B2026.03 | 81.8 | |
| EccLLM Backbone=Qwen2.5-7B2026.03 | 81.5 | |
| Eigen ScoreLLM Backbone=Qwen2.5-14B2026.03 | 81.5 | |
| EccLLM Backbone=Mistral-7B2026.03 | 81 | |
| EccLLM Backbone=Qwen2.5-3B2026.03 | 80.9 | |
| Eigen ScoreLLM Backbone=Mistral-7B2026.03 | 80.9 | |
| Eigen ScoreLLM Backbone=Llama-2-7B2026.03 | 80.7 | |
| Eigen ScoreLLM Backbone=Llama-3-8B2026.03 | 80.5 | |
| EccLLM Backbone=Qwen2.5-14B2026.03 | 80.2 | |
| EccLLM Backbone=Llama-2-7B2026.03 | 79.9 | |
| EccLLM Backbone=Llama-3-8B2026.03 | 79.2 | |
| ConULLM Backbone=Qwen2.5-7B2026.03 | 77.1 | |
| ConULLM Backbone=Mistral-7B2026.03 | 75.8 | |
| ConULLM Backbone=Qwen2.5-14B2026.03 | 75.4 | |
| ConULLM Backbone=Llama-3-8B2026.03 | 75.3 | |
| ConULLM Backbone=Llama-2-7B2026.03 | 75 | |
| ConULLM Backbone=Qwen2.5-3B2026.03 | 74.9 | |
| P(True)LLM Backbone=Qwen2.5-3B2026.03 | 73.9 | |
| P(True)LLM Backbone=Llama-3-8B2026.03 | 64 | |
| P(True)LLM Backbone=Mistral-7B2026.03 | 58.7 | |
| P(True)LLM Backbone=Llama-2-7B2026.03 | 53.3 |