Cross-lingual Retrieval on NeuCLIR 2023-2024 (test)
0.44MLIR (nDCG@20) 2023Qwen3-Embed
Evaluation Results
| Method | Links | |||
|---|---|---|---|---|
| Qwen3-Embedsize=4B2025.10 | 0.44 | 0.415 | 0.428 | |
| Qwen3-Embedsize=8B2025.10 | 0.434 | 0.419 | 0.427 | |
| SPLADE v3configuration=transl. docs2025.10 | 0.42 | 0.44 | 0.43 | |
| PLAID-X2025.10 | 0.404 | 0.468 | 0.436 | |
| MILCOtraining_components=SAP, SCTKD, LexEcho, parameters=560M2025.10 | 0.395 | 0.427 | 0.411 | |
| Qwen3-Embedsize=0.6B2025.10 | 0.317 | 0.311 | 0.314 |