RAG Performance Prediction on NQ-Open
0.793QE5 ScoreBert-Gen
Evaluation Results
| Method | Links | |||
|---|---|---|---|---|
| Bert-GenLLM=Llama, Retrieval Method=E5 Retrieval2026.04 | 0.793 | 0.597 | 0.502 | |
| Bert-GenLLM=Llama, Retrieval Method=BM25 Retrieval2026.04 | 0.788 | 0.661 | 0.451 | |
| Bert-GenLLM=Falcon, Retrieval Method=E5 Retrieval2026.04 | 0.765 | 0.62 | 0.611 | |
| Bert-GenLLM=Falcon, Retrieval Method=BM25 Retrieval2026.04 | 0.713 | 0.528 | 0.503 | |
| UncertaintyLLM=Falcon, Retrieval Method=E5 Retrieval2026.04 | 0.503 | 0.355 | 0.374 | |
| Bert-PostLLM=Falcon, Retrieval Method=E5 Retrieval2026.04 | 0.499 | 0.428 | 0.481 | |
| Bert-PostLLM=Falcon, Retrieval Method=BM25 Retrieval2026.04 | 0.45 | 0.285 | 0.428 | |
| Bert-PostLLM=Llama, Retrieval Method=BM25 Retrieval2026.04 | 0.447 | 0.292 | 0.312 | |
| UncertaintyLLM=Falcon, Retrieval Method=BM25 Retrieval2026.04 | 0.419 | 0.251 | 0.33 | |
| Bert-PostLLM=Llama, Retrieval Method=E5 Retrieval2026.04 | 0.39 | 0.21 | 0.244 | |
| UncertaintyLLM=Llama, Retrieval Method=BM25 Retrieval2026.04 | 0.354 | -0.022 | 0.168 | |
| UncertaintyLLM=Llama, Retrieval Method=E5 Retrieval2026.04 | 0.335 | 0.008 | 0.135 | |
| EntailmentLLM=Llama, Retrieval Method=BM25 Retrieval2026.04 | 0.317 | 0.135 | 0.158 | |
| REFLLM=Llama, Retrieval Method=BM25 Retrieval2026.04 | 0.305 | 0.19 | 0.226 | |
| REFLLM=Falcon, Retrieval Method=BM25 Retrieval2026.04 | 0.255 | 0.219 | 0.253 | |
| EntailmentLLM=Llama, Retrieval Method=E5 Retrieval2026.04 | 0.211 | 0.09 | 0.066 | |
| NQCLLM=Llama, Retrieval Method=E5 Retrieval2026.04 | 0.208 | 0.153 | 0.141 | |
| QCLLM=Llama, Retrieval Method=E5 Retrieval2026.04 | 0.204 | 0.15 | 0.138 | |
| SMVLLM=Llama, Retrieval Method=E5 Retrieval2026.04 | 0.204 | 0.148 | 0.136 | |
| U_SMVLLM=Llama, Retrieval Method=E5 Retrieval2026.04 | 0.201 | 0.146 | 0.134 | |
| QCLLM=Falcon, Retrieval Method=BM25 Retrieval2026.04 | 0.199 | 0.185 | 0.19 | |
| U_SMVLLM=Falcon, Retrieval Method=BM25 Retrieval2026.04 | 0.199 | 0.183 | 0.189 | |
| U_WIGLLM=Llama, Retrieval Method=E5 Retrieval2026.04 | 0.189 | 0.124 | 0.097 | |
| NQCLLM=Falcon, Retrieval Method=E5 Retrieval2026.04 | 0.167 | 0.171 | 0.203 | |
| U_WIGLLM=Llama, Retrieval Method=BM25 Retrieval2026.04 | 0.164 | 0.095 | 0.117 | |
| SMVLLM=Falcon, Retrieval Method=E5 Retrieval2026.04 | 0.163 | 0.164 | 0.196 | |
| EntailmentLLM=Falcon, Retrieval Method=BM25 Retrieval2026.04 | 0.159 | 0.135 | 0.152 | |
| WIGLLM=Falcon, Retrieval Method=BM25 Retrieval2026.04 | 0.158 | 0.149 | 0.136 | |
| WIGLLM=Llama, Retrieval Method=BM25 Retrieval2026.04 | 0.157 | 0.1 | 0.122 | |
| QCLLM=Falcon, Retrieval Method=E5 Retrieval2026.04 | 0.157 | 0.163 | 0.194 | |
| QCLLM=Llama, Retrieval Method=BM25 Retrieval2026.04 | 0.154 | 0.117 | 0.147 | |
| U_SMVLLM=Falcon, Retrieval Method=E5 Retrieval2026.04 | 0.154 | 0.156 | 0.187 | |
| U_SMVLLM=Llama, Retrieval Method=BM25 Retrieval2026.04 | 0.151 | 0.116 | 0.146 | |
| REFLLM=Falcon, Retrieval Method=E5 Retrieval2026.04 | 0.149 | 0.158 | 0.178 | |
| U_WIGLLM=Falcon, Retrieval Method=E5 Retrieval2026.04 | 0.142 | 0.133 | 0.157 | |
| meanIDFLLM=Falcon, Retrieval Method=BM25 Retrieval2026.04 | 0.141 | 0.121 | 0.12 | |
| U_WIGLLM=Falcon, Retrieval Method=BM25 Retrieval2026.04 | 0.141 | 0.136 | 0.131 | |
| maxIDFLLM=Falcon, Retrieval Method=BM25 Retrieval2026.04 | 0.128 | 0.108 | 0.121 | |
| meanVARLLM=Falcon, Retrieval Method=BM25 Retrieval2026.04 | 0.125 | 0.108 | 0.118 | |
| maxSCQLLM=Falcon, Retrieval Method=BM25 Retrieval2026.04 | 0.124 | 0.11 | 0.131 | |
| maxSCQLLM=Llama, Retrieval Method=BM25 Retrieval2026.04 | 0.122 | 0.079 | 0.099 | |
| meanSCQLLM=Falcon, Retrieval Method=BM25 Retrieval2026.04 | 0.12 | 0.11 | 0.108 | |
| maxIDFLLM=Llama, Retrieval Method=BM25 Retrieval2026.04 | 0.117 | 0.071 | 0.104 | |
| meanIDFLLM=Llama, Retrieval Method=BM25 Retrieval2026.04 | 0.109 | 0.084 | 0.103 | |
| NQCLLM=Falcon, Retrieval Method=BM25 Retrieval2026.04 | 0.108 | 0.081 | 0.083 | |
| SMVLLM=Falcon, Retrieval Method=BM25 Retrieval2026.04 | 0.108 | 0.081 | 0.083 | |
| REFLLM=Llama, Retrieval Method=E5 Retrieval2026.04 | 0.106 | 0.081 | 0.071 | |
| maxVARLLM=Llama, Retrieval Method=BM25 Retrieval2026.04 | 0.104 | 0.068 | 0.074 | |
| SMVLLM=Llama, Retrieval Method=BM25 Retrieval2026.04 | 0.1 | 0.067 | 0.087 | |
| WIGLLM=Llama, Retrieval Method=E5 Retrieval2026.04 | 0.099 | 0.071 | 0.067 | |
| meanVARLLM=Llama, Retrieval Method=BM25 Retrieval2026.04 | 0.098 | 0.071 | 0.08 | |
| NQCLLM=Llama, Retrieval Method=BM25 Retrieval2026.04 | 0.098 | 0.067 | 0.085 | |
| WIGLLM=Falcon, Retrieval Method=E5 Retrieval2026.04 | 0.097 | 0.072 | 0.084 | |
| meanSCQLLM=Llama, Retrieval Method=BM25 Retrieval2026.04 | 0.093 | 0.082 | 0.09 | |
| maxVARLLM=Falcon, Retrieval Method=BM25 Retrieval2026.04 | 0.093 | 0.083 | 0.1 | |
| EntailmentLLM=Falcon, Retrieval Method=E5 Retrieval2026.04 | 0.057 | 0.05 | 0.096 | |
| meanVARLLM=Falcon, Retrieval Method=E5 Retrieval2026.04 | 0.055 | 0.023 | 0.036 | |
| maxIDFLLM=Falcon, Retrieval Method=E5 Retrieval2026.04 | 0.053 | 0.038 | 0.041 | |
| maxSCQLLM=Falcon, Retrieval Method=E5 Retrieval2026.04 | 0.053 | 0.029 | 0.048 | |
| meanIDFLLM=Falcon, Retrieval Method=E5 Retrieval2026.04 | 0.046 | 0.015 | 0.016 | |
| maxVARLLM=Falcon, Retrieval Method=E5 Retrieval2026.04 | 0.036 | 0.014 | 0.03 | |
| meanSCQLLM=Falcon, Retrieval Method=E5 Retrieval2026.04 | 0.027 | -0.002 | 0.008 | |
| minVARLLM=Falcon, Retrieval Method=BM25 Retrieval2026.04 | 0.018 | -0.005 | -0.003 | |
| maxSCQLLM=Llama, Retrieval Method=E5 Retrieval2026.04 | 0.002 | -0.01 | -0.007 | |
| minVARLLM=Llama, Retrieval Method=BM25 Retrieval2026.04 | -0.001 | 0.017 | 0.015 | |
| maxIDFLLM=Llama, Retrieval Method=E5 Retrieval2026.04 | -0.008 | -0.008 | 0.004 | |
| minSCQLLM=Falcon, Retrieval Method=BM25 Retrieval2026.04 | -0.02 | -0.022 | -0.028 | |
| maxVARLLM=Llama, Retrieval Method=E5 Retrieval2026.04 | -0.021 | -0.019 | -0.01 | |
| minIDFLLM=Falcon, Retrieval Method=BM25 Retrieval2026.04 | -0.023 | -0.025 | -0.032 | |
| minVARLLM=Falcon, Retrieval Method=E5 Retrieval2026.04 | -0.023 | -0.054 | -0.042 | |
| minSCQLLM=Llama, Retrieval Method=BM25 Retrieval2026.04 | -0.026 | 0.002 | -0.014 | |
| minIDFLLM=Llama, Retrieval Method=BM25 Retrieval2026.04 | -0.027 | 0.006 | -0.013 | |
| meanVARLLM=Llama, Retrieval Method=E5 Retrieval2026.04 | -0.041 | -0.03 | -0.012 | |
| meanSCQLLM=Llama, Retrieval Method=E5 Retrieval2026.04 | -0.043 | -0.024 | -0.035 | |
| meanIDFLLM=Llama, Retrieval Method=E5 Retrieval2026.04 | -0.045 | -0.029 | -0.025 | |
| minVARLLM=Llama, Retrieval Method=E5 Retrieval2026.04 | -0.057 | -0.034 | -0.028 | |
| minSCQLLM=Llama, Retrieval Method=E5 Retrieval2026.04 | -0.059 | -0.039 | -0.053 | |
| minIDFLLM=Llama, Retrieval Method=E5 Retrieval2026.04 | -0.067 | -0.046 | -0.058 | |
| minSCQLLM=Falcon, Retrieval Method=E5 Retrieval2026.04 | -0.074 | -0.084 | -0.059 | |
| minIDFLLM=Falcon, Retrieval Method=E5 Retrieval2026.04 | -0.081 | -0.091 | -0.065 |