Information Retrieval on Academic Search Political Science
19.4MAP@100MiniLM_Houlsby
Evaluation results
| Method | Links | |||
|---|---|---|---|---|
| MiniLM_HoulsbyBackbone=MiniLM, Training Paradigm=Knowledge Distillation (Oracle), Adapter Type=Houlsby2026.02 | 19.4 | 59.2 | 30.7 | |
| MiniLM_HoulsbyBackbone=MiniLM, Training Paradigm=DRAMA, Adapter Type=Houlsby2026.02 | 19.4 | 59.2 | 30.7 | |
| MiniLM_LoRABackbone=MiniLM, Training Paradigm=Knowledge Distillation (Oracle), Adapter Type=LoRA2026.02 | 19.3 | 59.2 | 30.6 | |
| MiniLM_LoRABackbone=MiniLM, Training Paradigm=DRAMA, Adapter Type=LoRA2026.02 | 19.2 | 59.2 | 30.5 | |
| MiniLM_ALLBackbone=MiniLM, Training Paradigm=Full Model, Training Scope=ALL2026.02 | 19.1 | 59 | 30.4 | |
| MiniLM_SBackbone=MiniLM, Training Paradigm=Full Model, Training Scope=S2026.02 | 18.4 | 57.8 | 29.5 | |
| BM252026.02 | 13.3 | 50.2 | 24.1 | |
| DistilBERT-HoulsbyTraining Paradigm=DRAMA, Model Architecture=DistilBERT, Adapter Type=Houlsby2026.02 | 0.202 | 0.605 | 0.319 | |
| DistilBERT-HoulsbyTraining Paradigm=Knowledge Distillation (Oracle), Model Architecture=DistilBERT, Adapter Type=Houlsby2026.02 | 0.201 | 0.603 | 0.316 | |
| DistilBERT-LoRATraining Paradigm=Knowledge Distillation (Oracle), Model Architecture=DistilBERT, Adapter Type=LoRA2026.02 | 0.197 | 0.597 | 0.312 | |
| DistilBERT-LoRATraining Paradigm=DRAMA, Model Architecture=DistilBERT, Adapter Type=LoRA2026.02 | 0.196 | 0.596 | 0.312 | |
| DistilBERT-ALLTraining Paradigm=Full Model, Model Architecture=DistilBERT2026.02 | 0.191 | 0.589 | 0.306 | |
| BERT-TTraining Paradigm=Full Model, Model Architecture=BERT2026.02 | 0.19 | 0.589 | 0.303 | |
| DistilBERT-STraining Paradigm=Full Model, Model Architecture=DistilBERT2026.02 | 0.186 | 0.588 | 0.3 | |
| MiniLM_SArchitecture=CrossEncoder, Training Paradigm=Full Model, Model Variant=Specialized2026.02 | 0.184 | 0.584 | 0.296 | |
| DRAMA (MiniLM_Houlsby)Architecture=CrossEncoder, Training Paradigm=DRAMA, Adapter Type=Houlsby2026.02 | 0.182 | 0.584 | 0.296 | |
| MiniLM_HoulsbyArchitecture=CrossEncoder, Training Paradigm=Knowledge Distillation (Oracle), Adapter Type=Houlsby2026.02 | 0.181 | 0.584 | 0.296 | |
| MiniLM_ALLArchitecture=CrossEncoder, Training Paradigm=Full Model, Model Variant=General2026.02 | 0.177 | 0.579 | 0.292 | |
| MiniLM_LoRAArchitecture=CrossEncoder, Training Paradigm=Knowledge Distillation (Oracle), Adapter Type=LoRA2026.02 | 0.177 | 0.579 | 0.292 | |
| DRAMA (MiniLM_LoRA)Architecture=CrossEncoder, Training Paradigm=DRAMA, Adapter Type=LoRA2026.02 | 0.177 | 0.581 | 0.293 | |
| BM25Type=Unsupervised Baseline2026.02 | 0.133 | 0.502 | 0.241 | |
| BM25Training Paradigm=Lexical Baseline2026.02 | 0.133 | 0.502 | 0.241 |