conditions: # mDPR, tied encoders, pFT w/ MS MARCO, FT all - name: mdpr-tied-pft-msmarco-ft-all.ar eval_key: mrtydi-v1.1-arabic command: python -m pyserini.search.faiss --threads 16 --batch-size 512 --encoder-class auto --encoder castorini/mdpr-tied-pft-msmarco-ft-all --topics mrtydi-v1.1-arabic-${split} --index mrtydi-v1.1-arabic-mdpr-tied-pft-msmarco-ft-all --output $output --hits 100 splits: - split: train scores: - MRR@100: 0.9505 R@100: 1.0000 - split: dev scores: - MRR@100: 0.6913 R@100: 0.9165 - split: test scores: - MRR@100: 0.6949 R@100: 0.9004 - name: mdpr-tied-pft-msmarco-ft-all.bn eval_key: mrtydi-v1.1-bengali command: python -m pyserini.search.faiss --threads 16 --batch-size 512 --encoder-class auto --encoder castorini/mdpr-tied-pft-msmarco-ft-all --topics mrtydi-v1.1-bengali-${split} --index mrtydi-v1.1-bengali-mdpr-tied-pft-msmarco-ft-all --output $output --hits 100 splits: - split: train scores: - MRR@100: 0.9620 R@100: 1.0000 - split: dev scores: - MRR@100: 0.5897 R@100: 0.8977 - split: test scores: - MRR@100: 0.6228 R@100: 0.9550 - name: mdpr-tied-pft-msmarco-ft-all.en eval_key: mrtydi-v1.1-english command: python -m pyserini.search.faiss --threads 16 --batch-size 512 --encoder-class auto --encoder castorini/mdpr-tied-pft-msmarco-ft-all --topics mrtydi-v1.1-english-${split} --index mrtydi-v1.1-english-mdpr-tied-pft-msmarco-ft-all --output $output --hits 100 splits: - split: train scores: - MRR@100: 0.8278 R@100: 1.0000 - split: dev scores: - MRR@100: 0.5357 R@100: 0.8884 - split: test scores: - MRR@100: 0.4916 R@100: 0.8414 - name: mdpr-tied-pft-msmarco-ft-all.fi eval_key: mrtydi-v1.1-finnish command: python -m pyserini.search.faiss --threads 16 --batch-size 512 --encoder-class auto --encoder castorini/mdpr-tied-pft-msmarco-ft-all --topics mrtydi-v1.1-finnish-${split} --index mrtydi-v1.1-finnish-mdpr-tied-pft-msmarco-ft-all --output $output --hits 100 splits: - split: train scores: - MRR@100: 0.9577 R@100: 0.9997 - split: dev scores: - MRR@100: 0.6626 R@100: 0.9171 - split: test scores: - MRR@100: 0.5595 R@100: 0.8563 - name: mdpr-tied-pft-msmarco-ft-all.id eval_key: mrtydi-v1.1-indonesian command: python -m pyserini.search.faiss --threads 16 --batch-size 512 --encoder-class auto --encoder castorini/mdpr-tied-pft-msmarco-ft-all --topics mrtydi-v1.1-indonesian-${split} --index mrtydi-v1.1-indonesian-mdpr-tied-pft-msmarco-ft-all --output $output --hits 100 splits: - split: train scores: - MRR@100: 0.9469 R@100: 1.0000 - split: dev scores: - MRR@100: 0.6294 R@100: 0.9150 - split: test scores: - MRR@100: 0.5783 R@100: 0.8609 - name: mdpr-tied-pft-msmarco-ft-all.ja eval_key: mrtydi-v1.1-japanese command: python -m pyserini.search.faiss --threads 16 --batch-size 512 --encoder-class auto --encoder castorini/mdpr-tied-pft-msmarco-ft-all --topics mrtydi-v1.1-japanese-${split} --index mrtydi-v1.1-japanese-mdpr-tied-pft-msmarco-ft-all --output $output --hits 100 splits: - split: train scores: - MRR@100: 0.8802 R@100: 0.9997 - split: dev scores: - MRR@100: 0.5505 R@100: 0.8696 - split: test scores: - MRR@100: 0.5007 R@100: 0.8130 - name: mdpr-tied-pft-msmarco-ft-all.ko eval_key: mrtydi-v1.1-korean command: python -m pyserini.search.faiss --threads 16 --batch-size 512 --encoder-class auto --encoder castorini/mdpr-tied-pft-msmarco-ft-all --topics mrtydi-v1.1-korean-${split} --index mrtydi-v1.1-korean-mdpr-tied-pft-msmarco-ft-all --output $output --hits 100 splits: - split: train scores: - MRR@100: 0.9195 R@100: 1.0000 - split: dev scores: - MRR@100: 0.5645 R@100: 0.8663 - split: test scores: - MRR@100: 0.4861 R@100: 0.7854 - name: mdpr-tied-pft-msmarco-ft-all.ru eval_key: mrtydi-v1.1-russian command: python -m pyserini.search.faiss --threads 16 --batch-size 512 --encoder-class auto --encoder castorini/mdpr-tied-pft-msmarco-ft-all --topics mrtydi-v1.1-russian-${split} --index mrtydi-v1.1-russian-mdpr-tied-pft-msmarco-ft-all --output $output --hits 100 splits: - split: train scores: - MRR@100: 0.8473 R@100: 0.9994 - split: dev scores: - MRR@100: 0.5104 R@100: 0.8720 - split: test scores: - MRR@100: 0.5161 R@100: 0.8432 - name: mdpr-tied-pft-msmarco-ft-all.sw eval_key: mrtydi-v1.1-swahili command: python -m pyserini.search.faiss --threads 16 --batch-size 512 --encoder-class auto --encoder castorini/mdpr-tied-pft-msmarco-ft-all --topics mrtydi-v1.1-swahili-${split} --index mrtydi-v1.1-swahili-mdpr-tied-pft-msmarco-ft-all --output $output --hits 100 splits: - split: train scores: - MRR@100: 0.9515 R@100: 1.0000 - split: dev scores: - MRR@100: 0.6404 R@100: 0.9018 - split: test scores: - MRR@100: 0.6438 R@100: 0.8756 - name: mdpr-tied-pft-msmarco-ft-all.te eval_key: mrtydi-v1.1-telugu command: python -m pyserini.search.faiss --threads 16 --batch-size 512 --encoder-class auto --encoder castorini/mdpr-tied-pft-msmarco-ft-all --topics mrtydi-v1.1-telugu-${split} --index mrtydi-v1.1-telugu-mdpr-tied-pft-msmarco-ft-all --output $output --hits 100 splits: - split: train scores: - MRR@100: 0.9679 R@100: 1.0000 - split: dev scores: - MRR@100: 0.7962 R@100: 0.9593 - split: test scores: - MRR@100: 0.8908 R@100: 0.9659 - name: mdpr-tied-pft-msmarco-ft-all.th eval_key: mrtydi-v1.1-thai command: python -m pyserini.search.faiss --threads 16 --batch-size 512 --encoder-class auto --encoder castorini/mdpr-tied-pft-msmarco-ft-all --topics mrtydi-v1.1-thai-${split} --index mrtydi-v1.1-thai-mdpr-tied-pft-msmarco-ft-all --output $output --hits 100 splits: - split: train scores: - MRR@100: 0.9504 R@100: 1.0000 - split: dev scores: - MRR@100: 0.6670 R@100: 0.9114 - split: test scores: - MRR@100: 0.6175 R@100: 0.8826 # mDPR, tied encoders, pFT w/ MS MARCO - name: mdpr-tied-pft-msmarco.ar eval_key: mrtydi-v1.1-arabic command: python -m pyserini.search.faiss --threads 16 --batch-size 512 --encoder-class auto --encoder castorini/mdpr-tied-pft-msmarco --topics mrtydi-v1.1-arabic-${split} --index mrtydi-v1.1-arabic-mdpr-tied-pft-msmarco --output $output --hits 100 splits: - split: train scores: - MRR@100: 0.3957 R@100: 0.7818 - split: dev scores: - MRR@100: 0.3978 R@100: 0.7778 - split: test scores: - MRR@100: 0.4414 R@100: 0.7971 - name: mdpr-tied-pft-msmarco.bn eval_key: mrtydi-v1.1-bengali command: python -m pyserini.search.faiss --threads 16 --batch-size 512 --encoder-class auto --encoder castorini/mdpr-tied-pft-msmarco --topics mrtydi-v1.1-bengali-${split} --index mrtydi-v1.1-bengali-mdpr-tied-pft-msmarco --output $output --hits 100 splits: - split: train scores: - MRR@100: 0.2920 R@100: 0.7323 - split: dev scores: - MRR@100: 0.2993 R@100: 0.7318 - split: test scores: - MRR@100: 0.3969 R@100: 0.7838 - name: mdpr-tied-pft-msmarco.en eval_key: mrtydi-v1.1-english command: python -m pyserini.search.faiss --threads 16 --batch-size 512 --encoder-class auto --encoder castorini/mdpr-tied-pft-msmarco --topics mrtydi-v1.1-english-${split} --index mrtydi-v1.1-english-mdpr-tied-pft-msmarco --output $output --hits 100 splits: - split: train scores: - MRR@100: 0.3374 R@100: 0.8111 - split: dev scores: - MRR@100: 0.3451 R@100: 0.7995 - split: test scores: - MRR@100: 0.3270 R@100: 0.7536 - name: mdpr-tied-pft-msmarco.fi eval_key: mrtydi-v1.1-finnish command: python -m pyserini.search.faiss --threads 16 --batch-size 512 --encoder-class auto --encoder castorini/mdpr-tied-pft-msmarco --topics mrtydi-v1.1-finnish-${split} --index mrtydi-v1.1-finnish-mdpr-tied-pft-msmarco --output $output --hits 100 splits: - split: train scores: - MRR@100: 0.3668 R@100: 0.7337 - split: dev scores: - MRR@100: 0.3636 R@100: 0.7371 - split: test scores: - MRR@100: 0.2750 R@100: 0.6471 - name: mdpr-tied-pft-msmarco.id eval_key: mrtydi-v1.1-indonesian command: python -m pyserini.search.faiss --threads 16 --batch-size 512 --encoder-class auto --encoder castorini/mdpr-tied-pft-msmarco --topics mrtydi-v1.1-indonesian-${split} --index mrtydi-v1.1-indonesian-mdpr-tied-pft-msmarco --output $output --hits 100 splits: - split: train scores: - MRR@100: 0.2794 R@100: 0.7044 - split: dev scores: - MRR@100: 0.2853 R@100: 0.7198 - split: test scores: - MRR@100: 0.3520 R@100: 0.7356 - name: mdpr-tied-pft-msmarco.ja eval_key: mrtydi-v1.1-japanese command: python -m pyserini.search.faiss --threads 16 --batch-size 512 --encoder-class auto --encoder castorini/mdpr-tied-pft-msmarco --topics mrtydi-v1.1-japanese-${split} --index mrtydi-v1.1-japanese-mdpr-tied-pft-msmarco --output $output --hits 100 splits: - split: train scores: - MRR@100: 0.3089 R@100: 0.7603 - split: dev scores: - MRR@100: 0.3108 R@100: 0.7597 - split: test scores: - MRR@100: 0.3107 R@100: 0.7317 - name: mdpr-tied-pft-msmarco.ko eval_key: mrtydi-v1.1-korean command: python -m pyserini.search.faiss --threads 16 --batch-size 512 --encoder-class auto --encoder castorini/mdpr-tied-pft-msmarco --topics mrtydi-v1.1-korean-${split} --index mrtydi-v1.1-korean-mdpr-tied-pft-msmarco --output $output --hits 100 splits: - split: train scores: - MRR@100: 0.3003 R@100: 0.6907 - split: dev scores: - MRR@100: 0.3017 R@100: 0.7046 - split: test scores: - MRR@100: 0.2820 R@100: 0.6172 - name: mdpr-tied-pft-msmarco.ru eval_key: mrtydi-v1.1-russian command: python -m pyserini.search.faiss --threads 16 --batch-size 512 --encoder-class auto --encoder castorini/mdpr-tied-pft-msmarco --topics mrtydi-v1.1-russian-${split} --index mrtydi-v1.1-russian-mdpr-tied-pft-msmarco --output $output --hits 100 splits: - split: train scores: - MRR@100: 0.2856 R@100: 0.7305 - split: dev scores: - MRR@100: 0.2943 R@100: 0.7404 - split: test scores: - MRR@100: 0.3561 R@100: 0.7432 - name: mdpr-tied-pft-msmarco.sw eval_key: mrtydi-v1.1-swahili command: python -m pyserini.search.faiss --threads 16 --batch-size 512 --encoder-class auto --encoder castorini/mdpr-tied-pft-msmarco --topics mrtydi-v1.1-swahili-${split} --index mrtydi-v1.1-swahili-mdpr-tied-pft-msmarco --output $output --hits 100 splits: - split: train scores: - MRR@100: 0.2491 R@100: 0.5195 - split: dev scores: - MRR@100: 0.2447 R@100: 0.5266 - split: test scores: - MRR@100: 0.3418 R@100: 0.6343 - name: mdpr-tied-pft-msmarco.te eval_key: mrtydi-v1.1-telugu command: python -m pyserini.search.faiss --threads 16 --batch-size 512 --encoder-class auto --encoder castorini/mdpr-tied-pft-msmarco --topics mrtydi-v1.1-telugu-${split} --index mrtydi-v1.1-telugu-mdpr-tied-pft-msmarco --output $output --hits 100 splits: - split: train scores: - MRR@100: 0.3059 R@100: 0.7510 - split: dev scores: - MRR@100: 0.2995 R@100: 0.7355 - split: test scores: - MRR@100: 0.3102 R@100: 0.7817 - name: mdpr-tied-pft-msmarco.th eval_key: mrtydi-v1.1-thai command: python -m pyserini.search.faiss --threads 16 --batch-size 512 --encoder-class auto --encoder castorini/mdpr-tied-pft-msmarco --topics mrtydi-v1.1-thai-${split} --index mrtydi-v1.1-thai-mdpr-tied-pft-msmarco --output $output --hits 100 splits: - split: train scores: - MRR@100: 0.2334 R@100: 0.5851 - split: dev scores: - MRR@100: 0.2407 R@100: 0.5795 - split: test scores: - MRR@100: 0.2693 R@100: 0.5945 # mDPR, tied encoders, pFT w/ NQ - name: mdpr-tied-pft-nq.ar eval_key: mrtydi-v1.1-arabic command: python -m pyserini.search.faiss --threads 16 --batch-size 512 --encoder-class auto --encoder castorini/mdpr-tied-pft-nq --topics mrtydi-v1.1-arabic-${split} --index mrtydi-v1.1-arabic-mdpr-tied-pft-nq --output $output --hits 100 splits: - split: train scores: - MRR@100: 0.2087 R@100: 0.5854 - split: dev scores: - MRR@100: 0.2132 R@100: 0.5868 - split: test scores: - MRR@100: 0.2214 R@100: 0.6001 - name: mdpr-tied-pft-nq.bn eval_key: mrtydi-v1.1-bengali command: python -m pyserini.search.faiss --threads 16 --batch-size 512 --encoder-class auto --encoder castorini/mdpr-tied-pft-nq --topics mrtydi-v1.1-bengali-${split} --index mrtydi-v1.1-bengali-mdpr-tied-pft-nq --output $output --hits 100 splits: - split: train scores: - MRR@100: 0.2371 R@100: 0.6281 - split: dev scores: - MRR@100: 0.2414 R@100: 0.6409 - split: test scores: - MRR@100: 0.2535 R@100: 0.7072 - name: mdpr-tied-pft-nq.en eval_key: mrtydi-v1.1-english command: python -m pyserini.search.faiss --threads 16 --batch-size 512 --encoder-class auto --encoder castorini/mdpr-tied-pft-nq --topics mrtydi-v1.1-english-${split} --index mrtydi-v1.1-english-mdpr-tied-pft-nq --output $output --hits 100 splits: - split: train scores: - MRR@100: 0.2441 R@100: 0.7217 - split: dev scores: - MRR@100: 0.2359 R@100: 0.7187 - split: test scores: - MRR@100: 0.2433 R@100: 0.6893 - name: mdpr-tied-pft-nq.fi eval_key: mrtydi-v1.1-finnish command: python -m pyserini.search.faiss --threads 16 --batch-size 512 --encoder-class auto --encoder castorini/mdpr-tied-pft-nq --topics mrtydi-v1.1-finnish-${split} --index mrtydi-v1.1-finnish-mdpr-tied-pft-nq --output $output --hits 100 splits: - split: train scores: - MRR@100: 0.2996 R@100: 0.6787 - split: dev scores: - MRR@100: 0.3252 R@100: 0.7037 - split: test scores: - MRR@100: 0.2444 R@100: 0.6401 - name: mdpr-tied-pft-nq.id eval_key: mrtydi-v1.1-indonesian command: python -m pyserini.search.faiss --threads 16 --batch-size 512 --encoder-class auto --encoder castorini/mdpr-tied-pft-nq --topics mrtydi-v1.1-indonesian-${split} --index mrtydi-v1.1-indonesian-mdpr-tied-pft-nq --output $output --hits 100 splits: - split: train scores: - MRR@100: 0.2706 R@100: 0.7322 - split: dev scores: - MRR@100: 0.2719 R@100: 0.7394 - split: test scores: - MRR@100: 0.2815 R@100: 0.6914 - name: mdpr-tied-pft-nq.ja eval_key: mrtydi-v1.1-japanese command: python -m pyserini.search.faiss --threads 16 --batch-size 512 --encoder-class auto --encoder castorini/mdpr-tied-pft-nq --topics mrtydi-v1.1-japanese-${split} --index mrtydi-v1.1-japanese-mdpr-tied-pft-nq --output $output --hits 100 splits: - split: train scores: - MRR@100: 0.2165 R@100: 0.6043 - split: dev scores: - MRR@100: 0.2299 R@100: 0.6239 - split: test scores: - MRR@100: 0.2058 R@100: 0.5734 - name: mdpr-tied-pft-nq.ko eval_key: mrtydi-v1.1-korean command: python -m pyserini.search.faiss --threads 16 --batch-size 512 --encoder-class auto --encoder castorini/mdpr-tied-pft-nq --topics mrtydi-v1.1-korean-${split} --index mrtydi-v1.1-korean-mdpr-tied-pft-nq --output $output --hits 100 splits: - split: train scores: - MRR@100: 0.2527 R@100: 0.6556 - split: dev scores: - MRR@100: 0.2680 R@100: 0.6271 - split: test scores: - MRR@100: 0.2234 R@100: 0.5499 - name: mdpr-tied-pft-nq.ru eval_key: mrtydi-v1.1-russian command: python -m pyserini.search.faiss --threads 16 --batch-size 512 --encoder-class auto --encoder castorini/mdpr-tied-pft-nq --topics mrtydi-v1.1-russian-${split} --index mrtydi-v1.1-russian-mdpr-tied-pft-nq --output $output --hits 100 splits: - split: train scores: - MRR@100: 0.2160 R@100: 0.6262 - split: dev scores: - MRR@100: 0.2263 R@100: 0.6444 - split: test scores: - MRR@100: 0.2501 R@100: 0.6181 - name: mdpr-tied-pft-nq.sw eval_key: mrtydi-v1.1-swahili command: python -m pyserini.search.faiss --threads 16 --batch-size 512 --encoder-class auto --encoder castorini/mdpr-tied-pft-nq --topics mrtydi-v1.1-swahili-${split} --index mrtydi-v1.1-swahili-mdpr-tied-pft-nq --output $output --hits 100 splits: - split: train scores: - MRR@100: 0.2383 R@100: 0.5707 - split: dev scores: - MRR@100: 0.2543 R@100: 0.6138 - split: test scores: - MRR@100: 0.2621 R@100: 0.5965 - name: mdpr-tied-pft-nq.te eval_key: mrtydi-v1.1-telugu command: python -m pyserini.search.faiss --threads 16 --batch-size 512 --encoder-class auto --encoder castorini/mdpr-tied-pft-nq --topics mrtydi-v1.1-telugu-${split} --index mrtydi-v1.1-telugu-mdpr-tied-pft-nq --output $output --hits 100 splits: - split: train scores: - MRR@100: 0.1483 R@100: 0.4162 - split: dev scores: - MRR@100: 0.1494 R@100: 0.3967 - split: test scores: - MRR@100: 0.0970 R@100: 0.2454 - name: mdpr-tied-pft-nq.th eval_key: mrtydi-v1.1-thai command: python -m pyserini.search.faiss --threads 16 --batch-size 512 --encoder-class auto --encoder castorini/mdpr-tied-pft-nq --topics mrtydi-v1.1-thai-${split} --index mrtydi-v1.1-thai-mdpr-tied-pft-nq --output $output --hits 100 splits: - split: train scores: - MRR@100: 0.1426 R@100: 0.4717 - split: dev scores: - MRR@100: 0.1618 R@100: 0.4637 - split: test scores: - MRR@100: 0.1575 R@100: 0.4550 # mDPR, split encoders, pFT w/ NQ - name: mdpr-split-pft-nq.ar eval_key: mrtydi-v1.1-arabic command: python -m pyserini.search.faiss --threads 16 --batch-size 512 --encoder castorini/mdpr-question-nq --topics mrtydi-v1.1-arabic-${split} --index mrtydi-v1.1-arabic-mdpr-nq --output $output --hits 100 splits: - split: train scores: - MRR@100: 0.2510 R@100: 0.6384 - split: dev scores: - MRR@100: 0.2449 R@100: 0.6334 - split: test scores: - MRR@100: 0.2907 R@100: 0.6502 - name: mdpr-split-pft-nq.bn eval_key: mrtydi-v1.1-bengali command: python -m pyserini.search.faiss --threads 16 --batch-size 512 --encoder castorini/mdpr-question-nq --topics mrtydi-v1.1-bengali-${split} --index mrtydi-v1.1-bengali-mdpr-nq --output $output --hits 100 splits: - split: train scores: - MRR@100: 0.2293 R@100: 0.6454 - split: dev scores: - MRR@100: 0.2367 R@100: 0.6511 - split: test scores: - MRR@100: 0.2911 R@100: 0.7793 - name: mdpr-split-pft-nq.en eval_key: mrtydi-v1.1-english command: python -m pyserini.search.faiss --threads 16 --batch-size 512 --encoder castorini/mdpr-question-nq --topics mrtydi-v1.1-english-${split} --index mrtydi-v1.1-english-mdpr-nq --output $output --hits 100 splits: - split: train scores: - MRR@100: 0.2862 R@100: 0.7372 - split: dev scores: - MRR@100: 0.2821 R@100: 0.7437 - split: test scores: - MRR@100: 0.2907 R@100: 0.6779 - name: mdpr-split-pft-nq.fi eval_key: mrtydi-v1.1-finnish command: python -m pyserini.search.faiss --threads 16 --batch-size 512 --encoder castorini/mdpr-question-nq --topics mrtydi-v1.1-finnish-${split} --index mrtydi-v1.1-finnish-mdpr-nq --output $output --hits 100 splits: - split: train scores: - MRR@100: 0.2473 R@100: 0.6289 - split: dev scores: - MRR@100: 0.2466 R@100: 0.6283 - split: test scores: - MRR@100: 0.2050 R@100: 0.5680 - name: mdpr-split-pft-nq.id eval_key: mrtydi-v1.1-indonesian command: python -m pyserini.search.faiss --threads 16 --batch-size 512 --encoder castorini/mdpr-question-nq --topics mrtydi-v1.1-indonesian-${split} --index mrtydi-v1.1-indonesian-mdpr-nq --output $output --hits 100 splits: - split: train scores: - MRR@100: 0.2351 R@100: 0.6952 - split: dev scores: - MRR@100: 0.2475 R@100: 0.7181 - split: test scores: - MRR@100: 0.2705 R@100: 0.6848 - name: mdpr-split-pft-nq.ja eval_key: mrtydi-v1.1-japanese command: python -m pyserini.search.faiss --threads 16 --batch-size 512 --encoder castorini/mdpr-question-nq --topics mrtydi-v1.1-japanese-${split} --index mrtydi-v1.1-japanese-mdpr-nq --output $output --hits 100 splits: - split: train scores: - MRR@100: 0.1967 R@100: 0.5983 - split: dev scores: - MRR@100: 0.2055 R@100: 0.6142 - split: test scores: - MRR@100: 0.2119 R@100: 0.5840 - name: mdpr-split-pft-nq.ko eval_key: mrtydi-v1.1-korean command: python -m pyserini.search.faiss --threads 16 --batch-size 512 --encoder castorini/mdpr-question-nq --topics mrtydi-v1.1-korean-${split} --index mrtydi-v1.1-korean-mdpr-nq --output $output --hits 100 splits: - split: train scores: - MRR@100: 0.2383 R@100: 0.6180 - split: dev scores: - MRR@100: 0.2343 R@100: 0.6238 - split: test scores: - MRR@100: 0.2345 R@100: 0.5325 - name: mdpr-split-pft-nq.ru eval_key: mrtydi-v1.1-russian command: python -m pyserini.search.faiss --threads 16 --batch-size 512 --encoder castorini/mdpr-question-nq --topics mrtydi-v1.1-russian-${split} --index mrtydi-v1.1-russian-mdpr-nq --output $output --hits 100 splits: - split: train scores: - MRR@100: 0.2317 R@100: 0.6534 - split: dev scores: - MRR@100: 0.2490 R@100: 0.6553 - split: test scores: - MRR@100: 0.2820 R@100: 0.6474 - name: mdpr-split-pft-nq.sw eval_key: mrtydi-v1.1-swahili command: python -m pyserini.search.faiss --threads 16 --batch-size 512 --encoder castorini/mdpr-question-nq --topics mrtydi-v1.1-swahili-${split} --index mrtydi-v1.1-swahili-mdpr-nq --output $output --hits 100 splits: - split: train scores: - MRR@100: 0.1457 R@100: 0.4481 - split: dev scores: - MRR@100: 0.1547 R@100: 0.4724 - split: test scores: - MRR@100: 0.1883 R@100: 0.5281 - name: mdpr-split-pft-nq.te eval_key: mrtydi-v1.1-telugu command: python -m pyserini.search.faiss --threads 16 --batch-size 512 --encoder castorini/mdpr-question-nq --topics mrtydi-v1.1-telugu-${split} --index mrtydi-v1.1-telugu-mdpr-nq --output $output --hits 100 splits: - split: train scores: - MRR@100: 0.1489 R@100: 0.4905 - split: dev scores: - MRR@100: 0.1503 R@100: 0.4934 - split: test scores: - MRR@100: 0.1099 R@100: 0.3661 - name: mdpr-split-pft-nq.th eval_key: mrtydi-v1.1-thai command: python -m pyserini.search.faiss --threads 16 --batch-size 512 --encoder castorini/mdpr-question-nq --topics mrtydi-v1.1-thai-${split} --index mrtydi-v1.1-thai-mdpr-nq --output $output --hits 100 splits: - split: train scores: - MRR@100: 0.1603 R@100: 0.4983 - split: dev scores: - MRR@100: 0.1584 R@100: 0.5083 - split: test scores: - MRR@100: 0.1709 R@100: 0.5146 # BM25 - name: bm25.ar eval_key: mrtydi-v1.1-arabic command: python -m pyserini.search.lucene --threads 16 --batch-size 128 --language ar --topics mrtydi-v1.1-arabic-${split} --index mrtydi-v1.1-arabic --output $output --bm25 --hits 100 splits: - split: train scores: - MRR@100: 0.3356 R@100: 0.7944 - split: dev scores: - MRR@100: 0.3462 R@100: 0.7872 - split: test scores: - MRR@100: 0.3682 R@100: 0.7928 - name: bm25.bn eval_key: mrtydi-v1.1-bengali command: python -m pyserini.search.lucene --threads 16 --batch-size 128 --language bn --topics mrtydi-v1.1-bengali-${split} --index mrtydi-v1.1-bengali --output $output --bm25 --hits 100 splits: - split: train scores: - MRR@100: 0.3566 - R@100: 0.8336 - split: dev scores: - MRR@100: 0.3385 - R@100: 0.8432 - split: test scores: - MRR@100: 0.4182 - R@100: 0.8694 - name: bm25.en eval_key: mrtydi-v1.1-english command: python -m pyserini.search.lucene --threads 16 --batch-size 128 --language en --topics mrtydi-v1.1-english-${split} --index mrtydi-v1.1-english --output $output --bm25 --hits 100 splits: - split: train scores: - MRR@100: 0.1592 - R@100: 0.5785 - split: dev scores: - MRR@100: 0.1685 - R@100: 0.6196 - split: test scores: - MRR@100: 0.1404 - R@100: 0.5365 - name: bm25.fi eval_key: mrtydi-v1.1-finnish command: python -m pyserini.search.lucene --threads 16 --batch-size 128 --language fi --topics mrtydi-v1.1-finnish-${split} --index mrtydi-v1.1-finnish --output $output --bm25 --hits 100 splits: - split: train scores: - MRR@100: 0.4101 - R@100: 0.8198 - split: dev scores: - MRR@100: 0.4136 - R@100: 0.8285 - split: test scores: - MRR@100: 0.2836 - R@100: 0.7196 - name: bm25.id eval_key: mrtydi-v1.1-indonesian command: python -m pyserini.search.lucene --threads 16 --batch-size 128 --language id --topics mrtydi-v1.1-indonesian-${split} --index mrtydi-v1.1-indonesian --output $output --bm25 --hits 100 splits: - split: train scores: - MRR@100: 0.2972 - R@100: 0.7948 - split: dev scores: - MRR@100: 0.2937 - R@100: 0.7827 - split: test scores: - MRR@100: 0.3762 - R@100: 0.8426 - name: bm25.ja eval_key: mrtydi-v1.1-japanese command: python -m pyserini.search.lucene --threads 16 --batch-size 128 --language ja --topics mrtydi-v1.1-japanese-${split} --index mrtydi-v1.1-japanese --output $output --bm25 --hits 100 splits: - split: train scores: - MRR@100: 0.2262 - R@100: 0.7290 - split: dev scores: - MRR@100: 0.2250 - R@100: 0.7252 - split: test scores: - MRR@100: 0.2125 - R@100: 0.6431 - name: bm25.ko eval_key: mrtydi-v1.1-korean command: python -m pyserini.search.lucene --threads 16 --batch-size 128 --language ko --topics mrtydi-v1.1-korean-${split} --index mrtydi-v1.1-korean --output $output --bm25 --hits 100 splits: - split: train scores: - MRR@100: 0.2596 - R@100: 0.6178 - split: dev scores: - MRR@100: 0.2888 - R@100: 0.6733 - split: test scores: - MRR@100: 0.2848 - R@100: 0.6188 - name: bm25.ru eval_key: mrtydi-v1.1-russian command: python -m pyserini.search.lucene --threads 16 --batch-size 128 --language ru --topics mrtydi-v1.1-russian-${split} --index mrtydi-v1.1-russian --output $output --bm25 --hits 100 splits: - split: train scores: - MRR@100: 0.2229 - R@100: 0.5779 - split: dev scores: - MRR@100: 0.2202 - R@100: 0.5760 - split: test scores: - MRR@100: 0.3163 - R@100: 0.6541 - name: bm25.sw eval_key: mrtydi-v1.1-swahili command: python -m pyserini.search.lucene --threads 16 --batch-size 128 --language sw --topics mrtydi-v1.1-swahili-${split} --index mrtydi-v1.1-swahili --output $output --bm25 --hits 100 splits: - split: train scores: - MRR@100: 0.2610 - R@100: 0.5903 - split: dev scores: - MRR@100: 0.2693 - R@100: 0.5789 - split: test scores: - MRR@100: 0.3893 - R@100: 0.7642 - name: bm25.te eval_key: mrtydi-v1.1-telugu command: python -m pyserini.search.lucene --threads 16 --batch-size 128 --language te --topics mrtydi-v1.1-telugu-${split} --index mrtydi-v1.1-telugu --output $output --bm25 --hits 100 splits: - split: train scores: - MRR@100: 0.4204 - R@100: 0.8229 - split: dev scores: - MRR@100: 0.4269 - R@100: 0.8362 - split: test scores: - MRR@100: 0.5283 - R@100: 0.8971 - name: bm25.th eval_key: mrtydi-v1.1-thai command: python -m pyserini.search.lucene --threads 16 --batch-size 128 --language th --topics mrtydi-v1.1-thai-${split} --index mrtydi-v1.1-thai --output $output --bm25 --hits 100 splits: - split: train scores: - MRR@100: 0.3543 - R@100: 0.8349 - split: dev scores: - MRR@100: 0.3586 - R@100: 0.8536 - split: test scores: - MRR@100: 0.4012 - R@100: 0.8529