cat-claws/hotpotqa-variants

---
configs:
- config_name: all_mid_7
data_files:
- split: train
path: data/all_mid_7-train-*.parquet
- split: validation
path: data/all_mid_7-validation-*.parquet
- config_name: all_mid_7_llm
data_files:
- split: train
path: data/all_mid_7_llm-train-*.parquet
- split: validation
path: data/all_mid_7_llm-validation-*.parquet
- config_name: ambigattr
data_files:
- split: train
path: data/ambigattr-train-*.parquet
- split: validation
path: data/ambigattr-validation-*.parquet
- config_name: ambigattr_llm
data_files:
- split: train
path: data/ambigattr_llm-train-*.parquet
- split: validation
path: data/ambigattr_llm-validation-*.parquet
- config_name: ambigchoice
data_files:
- split: train
path: data/ambigchoice-train-*.parquet
- split: validation
path: data/ambigchoice-validation-*.parquet
- config_name: ambigchoice_llm
data_files:
- split: train
path: data/ambigchoice_llm-train-*.parquet
- split: validation
path: data/ambigchoice_llm-validation-*.parquet
- config_name: ambigref
data_files:
- split: train
path: data/ambigref-train-*.parquet
- split: validation
path: data/ambigref-validation-*.parquet
- config_name: ambigref_llm
data_files:
- split: train
path: data/ambigref_llm-train-*.parquet
- split: validation
path: data/ambigref_llm-validation-*.parquet
- config_name: annorel
data_files:
- split: train
path: data/annorel-train-*.parquet
- split: validation
path: data/annorel-validation-*.parquet
- config_name: annorel_llm
data_files:
- split: train
path: data/annorel_llm-train-*.parquet
- split: validation
path: data/annorel_llm-validation-*.parquet
- config_name: annotype
data_files:
- split: train
path: data/annotype-train-*.parquet
- split: validation
path: data/annotype-validation-*.parquet
- config_name: annotype_llm
data_files:
- split: train
path: data/annotype_llm-train-*.parquet
- split: validation
path: data/annotype_llm-validation-*.parquet
- config_name: ans_first_neg
data_files:
- split: train
path: data/ans_first_neg-train-*.parquet
- split: validation
path: data/ans_first_neg-validation-*.parquet
- config_name: ans_first_neg_llm
data_files:
- split: train
path: data/ans_first_neg_llm-train-*.parquet
- split: validation
path: data/ans_first_neg_llm-validation-*.parquet
- config_name: answer_then_detour
data_files:
- split: train
path: data/answer_then_detour-train-*.parquet
- split: validation
path: data/answer_then_detour-validation-*.parquet
- config_name: answer_then_detour_llm
data_files:
- split: train
path: data/answer_then_detour_llm-train-*.parquet
- split: validation
path: data/answer_then_detour_llm-validation-*.parquet
- config_name: approx
data_files:
- split: train
path: data/approx-train-*.parquet
- split: validation
path: data/approx-validation-*.parquet
- config_name: approx_llm
data_files:
- split: train
path: data/approx_llm-train-*.parquet
- split: validation
path: data/approx_llm-validation-*.parquet
- config_name: bae
data_files:
- split: train
path: data/bae-train-*.parquet
- split: validation
path: data/bae-validation-*.parquet
- config_name: bae_llm
data_files:
- split: train
path: data/bae_llm-train-*.parquet
- split: validation
path: data/bae_llm-validation-*.parquet
- config_name: behind
data_files:
- split: train
path: data/behind-train-*.parquet
- split: validation
path: data/behind-validation-*.parquet
- config_name: behind_llm
data_files:
- split: train
path: data/behind_llm-train-*.parquet
- split: validation
path: data/behind_llm-validation-*.parquet
- config_name: big_mixed
data_files:
- split: train
path: data/big_mixed-train-*.parquet
- split: validation
path: data/big_mixed-validation-*.parquet
- config_name: big_mixed_llm
data_files:
- split: train
path: data/big_mixed_llm-train-*.parquet
- split: validation
path: data/big_mixed_llm-validation-*.parquet
- config_name: bracketed
data_files:
- split: train
path: data/bracketed-train-*.parquet
- split: validation
path: data/bracketed-validation-*.parquet
- config_name: bracketed_llm
data_files:
- split: train
path: data/bracketed_llm-train-*.parquet
- split: validation
path: data/bracketed_llm-validation-*.parquet
- config_name: buried
data_files:
- split: train
path: data/buried-train-*.parquet
- split: validation
path: data/buried-validation-*.parquet
- config_name: buried_llm
data_files:
- split: train
path: data/buried_llm-train-*.parquet
- split: validation
path: data/buried_llm-validation-*.parquet
- config_name: buried_mid
data_files:
- split: train
path: data/buried_mid-train-*.parquet
- split: validation
path: data/buried_mid-validation-*.parquet
- config_name: buried_mid_llm
data_files:
- split: train
path: data/buried_mid_llm-train-*.parquet
- split: validation
path: data/buried_mid_llm-validation-*.parquet
- config_name: candidate_para_edit1
data_files:
- split: train
path: data/candidate_para_edit1-train-*.parquet
- split: validation
path: data/candidate_para_edit1-validation-*.parquet
- config_name: candidate_para_edit10_fast
data_files:
- split: train
path: data/candidate_para_edit10_fast-train-*.parquet
- split: validation
path: data/candidate_para_edit10_fast-validation-*.parquet
- config_name: candidate_para_edit11_fast
data_files:
- split: train
path: data/candidate_para_edit11_fast-train-*.parquet
- split: validation
path: data/candidate_para_edit11_fast-validation-*.parquet
- config_name: candidate_para_edit12_fast
data_files:
- split: train
path: data/candidate_para_edit12_fast-train-*.parquet
- split: validation
path: data/candidate_para_edit12_fast-validation-*.parquet
- config_name: candidate_para_edit13_fast
data_files:
- split: train
path: data/candidate_para_edit13_fast-train-*.parquet
- split: validation
path: data/candidate_para_edit13_fast-validation-*.parquet
- config_name: candidate_para_edit14_fast
data_files:
- split: train
path: data/candidate_para_edit14_fast-train-*.parquet
- split: validation
path: data/candidate_para_edit14_fast-validation-*.parquet
- config_name: candidate_para_edit2
data_files:
- split: train
path: data/candidate_para_edit2-train-*.parquet
- split: validation
path: data/candidate_para_edit2-validation-*.parquet
- config_name: ood13
data_files:
- split: train
path: data/ood13-train-*.parquet
- split: validation
path: data/ood13-validation-*.parquet
- config_name: candidate_para_edit4_fast
data_files:
- split: train
path: data/candidate_para_edit4_fast-train-*.parquet
- split: validation
path: data/candidate_para_edit4_fast-validation-*.parquet
- config_name: candidate_para_edit5_fast
data_files:
- split: train
path: data/candidate_para_edit5_fast-train-*.parquet
- split: validation
path: data/candidate_para_edit5_fast-validation-*.parquet
- config_name: candidate_para_edit6_fast
data_files:
- split: train
path: data/candidate_para_edit6_fast-train-*.parquet
- split: validation
path: data/candidate_para_edit6_fast-validation-*.parquet
- config_name: candidate_para_edit7_fast
data_files:
- split: train
path: data/candidate_para_edit7_fast-train-*.parquet
- split: validation
path: data/candidate_para_edit7_fast-validation-*.parquet
- config_name: candidate_para_edit8_fast
data_files:
- split: train
path: data/candidate_para_edit8_fast-train-*.parquet
- split: validation
path: data/candidate_para_edit8_fast-validation-*.parquet
- config_name: candidate_para_edit9_fast
data_files:
- split: train
path: data/candidate_para_edit9_fast-train-*.parquet
- split: validation
path: data/candidate_para_edit9_fast-validation-*.parquet
- config_name: candidate_para_scope_guarded
data_files:
- split: train
path: data/candidate_para_scope_guarded-train-*.parquet
- split: validation
path: data/candidate_para_scope_guarded-validation-*.parquet
- config_name: ood13_adaptive
data_files:
- split: train
path: data/ood13_adaptive-train-*.parquet
- split: validation
path: data/ood13_adaptive-validation-*.parquet
- config_name: cascade
data_files:
- split: train
path: data/cascade-train-*.parquet
- split: validation
path: data/cascade-validation-*.parquet
- config_name: cascade_llm
data_files:
- split: train
path: data/cascade_llm-train-*.parquet
- split: validation
path: data/cascade_llm-validation-*.parquet
- config_name: clare
data_files:
- split: train
path: data/clare-train-*.parquet
- split: validation
path: data/clare-validation-*.parquet
- config_name: clare_llm
data_files:
- split: train
path: data/clare_llm-train-*.parquet
- split: validation
path: data/clare_llm-validation-*.parquet
- config_name: clean
data_files:
- split: train
path: data/clean-train-*.parquet
- split: validation
path: data/clean-validation-*.parquet
- config_name: cluster
data_files:
- split: train
path: data/cluster-train-*.parquet
- split: validation
path: data/cluster-validation-*.parquet
- config_name: cluster_llm
data_files:
- split: train
path: data/cluster_llm-train-*.parquet
- split: validation
path: data/cluster_llm-validation-*.parquet
- config_name: concessive
data_files:
- split: train
path: data/concessive-train-*.parquet
- split: validation
path: data/concessive-validation-*.parquet
- config_name: concessive_llm
data_files:
- split: train
path: data/concessive_llm-train-*.parquet
- split: validation
path: data/concessive_llm-validation-*.parquet
- config_name: consensus_then
data_files:
- split: train
path: data/consensus_then-train-*.parquet
- split: validation
path: data/consensus_then-validation-*.parquet
- config_name: consensus_then_llm
data_files:
- split: train
path: data/consensus_then_llm-train-*.parquet
- split: validation
path: data/consensus_then_llm-validation-*.parquet
- config_name: contrastive_afterthought
data_files:
- split: train
path: data/contrastive_afterthought-train-*.parquet
- split: validation
path: data/contrastive_afterthought-validation-*.parquet
- config_name: contrastive_afterthought_llm
data_files:
- split: train
path: data/contrastive_afterthought_llm-train-*.parquet
- split: validation
path: data/contrastive_afterthought_llm-validation-*.parquet
- config_name: cross_buried
data_files:
- split: train
path: data/cross_buried-train-*.parquet
- split: validation
path: data/cross_buried-validation-*.parquet
- config_name: cross_buried_llm
data_files:
- split: train
path: data/cross_buried_llm-train-*.parquet
- split: validation
path: data/cross_buried_llm-validation-*.parquet
- config_name: cross_five
data_files:
- split: train
path: data/cross_five-train-*.parquet
- split: validation
path: data/cross_five-validation-*.parquet
- config_name: cross_five_llm
data_files:
- split: train
path: data/cross_five_llm-train-*.parquet
- split: validation
path: data/cross_five_llm-validation-*.parquet
- config_name: cross_midlist
data_files:
- split: train
path: data/cross_midlist-train-*.parquet
- split: validation
path: data/cross_midlist-validation-*.parquet
- config_name: cross_midlist_llm
data_files:
- split: train
path: data/cross_midlist_llm-train-*.parquet
- split: validation
path: data/cross_midlist_llm-validation-*.parquet
- config_name: cross_shadow
data_files:
- split: train
path: data/cross_shadow-train-*.parquet
- split: validation
path: data/cross_shadow-validation-*.parquet
- config_name: cross_shadow_llm
data_files:
- split: train
path: data/cross_shadow_llm-train-*.parquet
- split: validation
path: data/cross_shadow_llm-validation-*.parquet
- config_name: cross_triple
data_files:
- split: train
path: data/cross_triple-train-*.parquet
- split: validation
path: data/cross_triple-validation-*.parquet
- config_name: cross_triple_llm
data_files:
- split: train
path: data/cross_triple_llm-train-*.parquet
- split: validation
path: data/cross_triple_llm-validation-*.parquet
- config_name: doubted
data_files:
- split: train
path: data/doubted-train-*.parquet
- split: validation
path: data/doubted-validation-*.parquet
- config_name: doubted_llm
data_files:
- split: train
path: data/doubted_llm-train-*.parquet
- split: validation
path: data/doubted_llm-validation-*.parquet
- config_name: downplayed
data_files:
- split: train
path: data/downplayed-train-*.parquet
- split: validation
path: data/downplayed-validation-*.parquet
- config_name: downplayed_llm
data_files:
- split: train
path: data/downplayed_llm-train-*.parquet
- split: validation
path: data/downplayed_llm-validation-*.parquet
- config_name: dual_shadow
data_files:
- split: train
path: data/dual_shadow-train-*.parquet
- split: validation
path: data/dual_shadow-validation-*.parquet
- config_name: dual_shadow_llm
data_files:
- split: train
path: data/dual_shadow_llm-train-*.parquet
- split: validation
path: data/dual_shadow_llm-validation-*.parquet
- config_name: ellipsis_list
data_files:
- split: train
path: data/ellipsis_list-train-*.parquet
- split: validation
path: data/ellipsis_list-validation-*.parquet
- config_name: ellipsis_list_llm
data_files:
- split: train
path: data/ellipsis_list_llm-train-*.parquet
- split: validation
path: data/ellipsis_list_llm-validation-*.parquet
- config_name: epistemic
data_files:
- split: train
path: data/epistemic-train-*.parquet
- split: validation
path: data/epistemic-validation-*.parquet
- config_name: epistemic_llm
data_files:
- split: train
path: data/epistemic_llm-train-*.parquet
- split: validation
path: data/epistemic_llm-validation-*.parquet
- config_name: expected
data_files:
- split: train
path: data/expected-train-*.parquet
- split: validation
path: data/expected-validation-*.parquet
- config_name: expected_llm
data_files:
- split: train
path: data/expected_llm-train-*.parquet
- split: validation
path: data/expected_llm-validation-*.parquet
- config_name: faster_ga
data_files:
- split: train
path: data/faster_ga-train-*.parquet
- split: validation
path: data/faster_ga-validation-*.parquet
- config_name: faster_ga_llm
data_files:
- split: train
path: data/faster_ga_llm-train-*.parquet
- split: validation
path: data/faster_ga_llm-validation-*.parquet
- config_name: feihua
data_files:
- split: train
path: data/feihua-train-*.parquet
- split: validation
path: data/feihua-validation-*.parquet
- config_name: feihua_llm
data_files:
- split: train
path: data/feihua_llm-train-*.parquet
- split: validation
path: data/feihua_llm-validation-*.parquet
- config_name: five_endblocked
data_files:
- split: train
path: data/five_endblocked-train-*.parquet
- split: validation
path: data/five_endblocked-validation-*.parquet
- config_name: five_endblocked_llm
data_files:
- split: train
path: data/five_endblocked_llm-train-*.parquet
- split: validation
path: data/five_endblocked_llm-validation-*.parquet
- config_name: five_item_mid
data_files:
- split: train
path: data/five_item_mid-train-*.parquet
- split: validation
path: data/five_item_mid-validation-*.parquet
- config_name: five_item_mid_llm
data_files:
- split: train
path: data/five_item_mid_llm-train-*.parquet
- split: validation
path: data/five_item_mid_llm-validation-*.parquet
- config_name: five_shadow
data_files:
- split: train
path: data/five_shadow-train-*.parquet
- split: validation
path: data/five_shadow-validation-*.parquet
- config_name: five_shadow_llm
data_files:
- split: train
path: data/five_shadow_llm-train-*.parquet
- split: validation
path: data/five_shadow_llm-validation-*.parquet
- config_name: footnote
data_files:
- split: train
path: data/footnote-train-*.parquet
- split: validation
path: data/footnote-validation-*.parquet
- config_name: footnote_llm
data_files:
- split: train
path: data/footnote_llm-train-*.parquet
- split: validation
path: data/footnote_llm-validation-*.parquet
- config_name: four_item
data_files:
- split: train
path: data/four_item-train-*.parquet
- split: validation
path: data/four_item-validation-*.parquet
- config_name: four_item_llm
data_files:
- split: train
path: data/four_item_llm-train-*.parquet
- split: validation
path: data/four_item_llm-validation-*.parquet
- config_name: giant_mixed
data_files:
- split: train
path: data/giant_mixed-train-*.parquet
- split: validation
path: data/giant_mixed-validation-*.parquet
- config_name: giant_mixed_llm
data_files:
- split: train
path: data/giant_mixed_llm-train-*.parquet
- split: validation
path: data/giant_mixed_llm-validation-*.parquet
- config_name: hbefore
data_files:
- split: train
path: data/hbefore-train-*.parquet
- split: validation
path: data/hbefore-validation-*.parquet
- config_name: hbefore_llm
data_files:
- split: train
path: data/hbefore_llm-train-*.parquet
- split: validation
path: data/hbefore_llm-validation-*.parquet
- config_name: hedged_list
data_files:
- split: train
path: data/hedged_list-train-*.parquet
- split: validation
path: data/hedged_list-validation-*.parquet
- config_name: hedged_list_llm
data_files:
- split: train
path: data/hedged_list_llm-train-*.parquet
- split: validation
path: data/hedged_list_llm-validation-*.parquet
- config_name: homoglyph
data_files:
- split: train
path: data/homoglyph-train-*.parquet
- split: validation
path: data/homoglyph-validation-*.parquet
- config_name: homoglyph_llm
data_files:
- split: train
path: data/homoglyph_llm-train-*.parquet
- split: validation
path: data/homoglyph_llm-validation-*.parquet
- config_name: hotflip
data_files:
- split: train
path: data/hotflip-train-*.parquet
- split: validation
path: data/hotflip-validation-*.parquet
- config_name: hotflip_llm
data_files:
- split: train
path: data/hotflip_llm-train-*.parquet
- split: validation
path: data/hotflip_llm-validation-*.parquet
- config_name: howeverafter
data_files:
- split: train
path: data/howeverafter-train-*.parquet
- split: validation
path: data/howeverafter-validation-*.parquet
- config_name: howeverafter_llm
data_files:
- split: train
path: data/howeverafter_llm-train-*.parquet
- split: validation
path: data/howeverafter_llm-validation-*.parquet
- config_name: hybrid_35
data_files:
- split: train
path: data/hybrid_35-train-*.parquet
- split: validation
path: data/hybrid_35-validation-*.parquet
- config_name: hybrid_35_llm
data_files:
- split: train
path: data/hybrid_35_llm-train-*.parquet
- split: validation
path: data/hybrid_35_llm-validation-*.parquet
- config_name: hybrid_mid_7_10
data_files:
- split: train
path: data/hybrid_mid_7_10-train-*.parquet
- split: validation
path: data/hybrid_mid_7_10-validation-*.parquet
- config_name: hybrid_mid_7_10_llm
data_files:
- split: train
path: data/hybrid_mid_7_10_llm-train-*.parquet
- split: validation
path: data/hybrid_mid_7_10_llm-validation-*.parquet
- config_name: hypernym
data_files:
- split: train
path: data/hypernym-train-*.parquet
- split: validation
path: data/hypernym-validation-*.parquet
- config_name: hypernym_llm
data_files:
- split: train
path: data/hypernym_llm-train-*.parquet
- split: validation
path: data/hypernym_llm-validation-*.parquet
- config_name: interleaved
data_files:
- split: train
path: data/interleaved-train-*.parquet
- split: validation
path: data/interleaved-validation-*.parquet
- config_name: interleaved_llm
data_files:
- split: train
path: data/interleaved_llm-train-*.parquet
- split: validation
path: data/interleaved_llm-validation-*.parquet
- config_name: inverted
data_files:
- split: train
path: data/inverted-train-*.parquet
- split: validation
path: data/inverted-validation-*.parquet
- config_name: inverted_llm
data_files:
- split: train
path: data/inverted_llm-train-*.parquet
- split: validation
path: data/inverted_llm-validation-*.parquet
- config_name: long_buried
data_files:
- split: train
path: data/long_buried-train-*.parquet
- split: validation
path: data/long_buried-validation-*.parquet
- config_name: long_buried_llm
data_files:
- split: train
path: data/long_buried_llm-train-*.parquet
- split: validation
path: data/long_buried_llm-validation-*.parquet
- config_name: long_shadow
data_files:
- split: train
path: data/long_shadow-train-*.parquet
- split: validation
path: data/long_shadow-validation-*.parquet
- config_name: long_shadow_llm
data_files:
- split: train
path: data/long_shadow_llm-train-*.parquet
- split: validation
path: data/long_shadow_llm-validation-*.parquet
- config_name: loosely
data_files:
- split: train
path: data/loosely-train-*.parquet
- split: validation
path: data/loosely-validation-*.parquet
- config_name: loosely_llm
data_files:
- split: train
path: data/loosely_llm-train-*.parquet
- split: validation
path: data/loosely_llm-validation-*.parquet
- config_name: mega_giant
data_files:
- split: train
path: data/mega_giant-train-*.parquet
- split: validation
path: data/mega_giant-validation-*.parquet
- config_name: mega_giant_llm
data_files:
- split: train
path: data/mega_giant_llm-train-*.parquet
- split: validation
path: data/mega_giant_llm-validation-*.parquet
- config_name: meta_wrong
data_files:
- split: train
path: data/meta_wrong-train-*.parquet
- split: validation
path: data/meta_wrong-validation-*.parquet
- config_name: meta_wrong_llm
data_files:
- split: train
path: data/meta_wrong_llm-train-*.parquet
- split: validation
path: data/meta_wrong_llm-validation-*.parquet
- config_name: mid_clause_anchor
data_files:
- split: train
path: data/mid_clause_anchor-train-*.parquet
- split: validation
path: data/mid_clause_anchor-validation-*.parquet
- config_name: mid_clause_anchor_llm
data_files:
- split: train
path: data/mid_clause_anchor_llm-train-*.parquet
- split: validation
path: data/mid_clause_anchor_llm-validation-*.parquet
- config_name: midlist
data_files:
- split: train
path: data/midlist-train-*.parquet
- split: validation
path: data/midlist-validation-*.parquet
- config_name: midlist_llm
data_files:
- split: train
path: data/midlist_llm-train-*.parquet
- split: validation
path: data/midlist_llm-validation-*.parquet
- config_name: midlist_long
data_files:
- split: train
path: data/midlist_long-train-*.parquet
- split: validation
path: data/midlist_long-validation-*.parquet
- config_name: midlist_long_llm
data_files:
- split: train
path: data/midlist_long_llm-train-*.parquet
- split: validation
path: data/midlist_long_llm-validation-*.parquet
- config_name: mirrored
data_files:
- split: train
path: data/mirrored-train-*.parquet
- split: validation
path: data/mirrored-validation-*.parquet
- config_name: mirrored_llm
data_files:
- split: train
path: data/mirrored_llm-train-*.parquet
- split: validation
path: data/mirrored_llm-validation-*.parquet
- config_name: misframe
data_files:
- split: train
path: data/misframe-train-*.parquet
- split: validation
path: data/misframe-validation-*.parquet
- config_name: misleads
data_files:
- split: train
path: data/misleads-train-*.parquet
- split: validation
path: data/misleads-validation-*.parquet
- config_name: misleads_llm
data_files:
- split: train
path: data/misleads_llm-train-*.parquet
- split: validation
path: data/misleads_llm-validation-*.parquet
- config_name: mixed_10_noise
data_files:
- split: train
path: data/mixed_10_noise-train-*.parquet
- split: validation
path: data/mixed_10_noise-validation-*.parquet
- config_name: mixed_10_noise_llm
data_files:
- split: train
path: data/mixed_10_noise_llm-train-*.parquet
- split: validation
path: data/mixed_10_noise_llm-validation-*.parquet
- config_name: mixed_all
data_files:
- split: train
path: data/mixed_all-train-*.parquet
- split: validation
path: data/mixed_all-validation-*.parquet
- config_name: mixed_all_llm
data_files:
- split: train
path: data/mixed_all_llm-train-*.parquet
- split: validation
path: data/mixed_all_llm-validation-*.parquet
- config_name: mixed_best
data_files:
- split: train
path: data/mixed_best-train-*.parquet
- split: validation
path: data/mixed_best-validation-*.parquet
- config_name: mixed_best_llm
data_files:
- split: train
path: data/mixed_best_llm-train-*.parquet
- split: validation
path: data/mixed_best_llm-validation-*.parquet
- config_name: mixed_extended
data_files:
- split: train
path: data/mixed_extended-train-*.parquet
- split: validation
path: data/mixed_extended-validation-*.parquet
- config_name: mixed_extended_llm
data_files:
- split: train
path: data/mixed_extended_llm-train-*.parquet
- split: validation
path: data/mixed_extended_llm-validation-*.parquet
- config_name: mixed_pos
data_files:
- split: train
path: data/mixed_pos-train-*.parquet
- split: validation
path: data/mixed_pos-validation-*.parquet
- config_name: mixed_pos_llm
data_files:
- split: train
path: data/mixed_pos_llm-train-*.parquet
- split: validation
path: data/mixed_pos_llm-validation-*.parquet
- config_name: negation
data_files:
- split: train
path: data/negation-train-*.parquet
- split: validation
path: data/negation-validation-*.parquet
- config_name: negation_llm
data_files:
- split: train
path: data/negation_llm-train-*.parquet
- split: validation
path: data/negation_llm-validation-*.parquet
- config_name: nested_gen
data_files:
- split: train
path: data/nested_gen-train-*.parquet
- split: validation
path: data/nested_gen-validation-*.parquet
- config_name: nested_gen_llm
data_files:
- split: train
path: data/nested_gen_llm-train-*.parquet
- split: validation
path: data/nested_gen_llm-validation-*.parquet
- config_name: open_list
data_files:
- split: train
path: data/open_list-train-*.parquet
- split: validation
path: data/open_list-validation-*.parquet
- config_name: open_list_llm
data_files:
- split: train
path: data/open_list_llm-train-*.parquet
- split: validation
path: data/open_list_llm-validation-*.parquet
- config_name: overridden
data_files:
- split: train
path: data/overridden-train-*.parquet
- split: validation
path: data/overridden-validation-*.parquet
- config_name: overridden_llm
data_files:
- split: train
path: data/overridden_llm-train-*.parquet
- split: validation
path: data/overridden_llm-validation-*.parquet
- config_name: para
data_files:
- split: train
path: data/para-train-*.parquet
- split: validation
path: data/para-validation-*.parquet
- config_name: para_llm
data_files:
- split: train
path: data/para_llm-train-*.parquet
- split: validation
path: data/para_llm-validation-*.parquet
- config_name: paraf
data_files:
- split: train
path: data/paraf-train-*.parquet
- split: validation
path: data/paraf-validation-*.parquet
- config_name: paraf_llm
data_files:
- split: train
path: data/paraf_llm-train-*.parquet
- split: validation
path: data/paraf_llm-validation-*.parquet
- config_name: paren
data_files:
- split: train
path: data/paren-train-*.parquet
- split: validation
path: data/paren-validation-*.parquet
- config_name: paren_llm
data_files:
- split: train
path: data/paren_llm-train-*.parquet
- split: validation
path: data/paren_llm-validation-*.parquet
- config_name: perturbed
data_files:
- split: train
path: data/perturbed-train-*.parquet
- split: validation
path: data/perturbed-validation-*.parquet
- config_name: perturbed_llm
data_files:
- split: train
path: data/perturbed_llm-train-*.parquet
- split: validation
path: data/perturbed_llm-validation-*.parquet
- config_name: pos2345_of_7
data_files:
- split: train
path: data/pos2345_of_7-train-*.parquet
- split: validation
path: data/pos2345_of_7-validation-*.parquet
- config_name: pos2345_of_7_llm
data_files:
- split: train
path: data/pos2345_of_7_llm-train-*.parquet
- split: validation
path: data/pos2345_of_7_llm-validation-*.parquet
- config_name: pos234_of_10
data_files:
- split: train
path: data/pos234_of_10-train-*.parquet
- split: validation
path: data/pos234_of_10-validation-*.parquet
- config_name: pos234_of_10_llm
data_files:
- split: train
path: data/pos234_of_10_llm-train-*.parquet
- split: validation
path: data/pos234_of_10_llm-validation-*.parquet
- config_name: pos234_of_15
data_files:
- split: train
path: data/pos234_of_15-train-*.parquet
- split: validation
path: data/pos234_of_15-validation-*.parquet
- config_name: pos234_of_15_llm
data_files:
- split: train
path: data/pos234_of_15_llm-train-*.parquet
- split: validation
path: data/pos234_of_15_llm-validation-*.parquet
- config_name: pos234_of_6
data_files:
- split: train
path: data/pos234_of_6-train-*.parquet
- split: validation
path: data/pos234_of_6-validation-*.parquet
- config_name: pos234_of_6_llm
data_files:
- split: train
path: data/pos234_of_6_llm-train-*.parquet
- split: validation
path: data/pos234_of_6_llm-validation-*.parquet
- config_name: pos23_of_10
data_files:
- split: train
path: data/pos23_of_10-train-*.parquet
- split: validation
path: data/pos23_of_10-validation-*.parquet
- config_name: pos23_of_10_llm
data_files:
- split: train
path: data/pos23_of_10_llm-train-*.parquet
- split: validation
path: data/pos23_of_10_llm-validation-*.parquet
- config_name: pos2_of_10
data_files:
- split: train
path: data/pos2_of_10-train-*.parquet
- split: validation
path: data/pos2_of_10-validation-*.parquet
- config_name: pos2_of_10_llm
data_files:
- split: train
path: data/pos2_of_10_llm-train-*.parquet
- split: validation
path: data/pos2_of_10_llm-validation-*.parquet
- config_name: pos2_of_15
data_files:
- split: train
path: data/pos2_of_15-train-*.parquet
- split: validation
path: data/pos2_of_15-validation-*.parquet
- config_name: pos2_of_15_llm
data_files:
- split: train
path: data/pos2_of_15_llm-train-*.parquet
- split: validation
path: data/pos2_of_15_llm-validation-*.parquet
- config_name: pos2_of_5
data_files:
- split: train
path: data/pos2_of_5-train-*.parquet
- split: validation
path: data/pos2_of_5-validation-*.parquet
- config_name: pos2_of_5_llm
data_files:
- split: train
path: data/pos2_of_5_llm-train-*.parquet
- split: validation
path: data/pos2_of_5_llm-validation-*.parquet
- config_name: pos2_of_6
data_files:
- split: train
path: data/pos2_of_6-train-*.parquet
- split: validation
path: data/pos2_of_6-validation-*.parquet
- config_name: pos2_of_6_llm
data_files:
- split: train
path: data/pos2_of_6_llm-train-*.parquet
- split: validation
path: data/pos2_of_6_llm-validation-*.parquet
- config_name: pos2to9_of_10
data_files:
- split: train
path: data/pos2to9_of_10-train-*.parquet
- split: validation
path: data/pos2to9_of_10-validation-*.parquet
- config_name: pos2to9_of_10_llm
data_files:
- split: train
path: data/pos2to9_of_10_llm-train-*.parquet
- split: validation
path: data/pos2to9_of_10_llm-validation-*.parquet
- config_name: pos345_of_7
data_files:
- split: train
path: data/pos345_of_7-train-*.parquet
- split: validation
path: data/pos345_of_7-validation-*.parquet
- config_name: pos345_of_7_llm
data_files:
- split: train
path: data/pos345_of_7_llm-train-*.parquet
- split: validation
path: data/pos345_of_7_llm-validation-*.parquet
- config_name: pos3_of_5
data_files:
- split: train
path: data/pos3_of_5-train-*.parquet
- split: validation
path: data/pos3_of_5-validation-*.parquet
- config_name: pos3_of_5_llm
data_files:
- split: train
path: data/pos3_of_5_llm-train-*.parquet
- split: validation
path: data/pos3_of_5_llm-validation-*.parquet
- config_name: pos3_of_7
data_files:
- split: train
path: data/pos3_of_7-train-*.parquet
- split: validation
path: data/pos3_of_7-validation-*.parquet
- config_name: pos3_of_7_llm
data_files:
- split: train
path: data/pos3_of_7_llm-train-*.parquet
- split: validation
path: data/pos3_of_7_llm-validation-*.parquet
- config_name: pos456_of_10
data_files:
- split: train
path: data/pos456_of_10-train-*.parquet
- split: validation
path: data/pos456_of_10-validation-*.parquet
- config_name: pos456_of_10_llm
data_files:
- split: train
path: data/pos456_of_10_llm-train-*.parquet
- split: validation
path: data/pos456_of_10_llm-validation-*.parquet
- config_name: pos5_of_10
data_files:
- split: train
path: data/pos5_of_10-train-*.parquet
- split: validation
path: data/pos5_of_10-validation-*.parquet
- config_name: pos5_of_10_llm
data_files:
- split: train
path: data/pos5_of_10_llm-train-*.parquet
- split: validation
path: data/pos5_of_10_llm-validation-*.parquet
- config_name: prefix_actually
data_files:
- split: train
path: data/prefix_actually-train-*.parquet
- split: validation
path: data/prefix_actually-validation-*.parquet
- config_name: prefix_actually_llm
data_files:
- split: train
path: data/prefix_actually_llm-train-*.parquet
- split: validation
path: data/prefix_actually_llm-validation-*.parquet
- config_name: prefix_theanswer
data_files:
- split: train
path: data/prefix_theanswer-train-*.parquet
- split: validation
path: data/prefix_theanswer-validation-*.parquet
- config_name: prefix_theanswer_llm
data_files:
- split: train
path: data/prefix_theanswer_llm-train-*.parquet
- split: validation
path: data/prefix_theanswer_llm-validation-*.parquet
- config_name: prefix_well
data_files:
- split: train
path: data/prefix_well-train-*.parquet
- split: validation
path: data/prefix_well-validation-*.parquet
- config_name: prefix_well_llm
data_files:
- split: train
path: data/prefix_well_llm-train-*.parquet
- split: validation
path: data/prefix_well_llm-validation-*.parquet
- config_name: primarily
data_files:
- split: train
path: data/primarily-train-*.parquet
- split: validation
path: data/primarily-validation-*.parquet
- config_name: primarily_llm
data_files:
- split: train
path: data/primarily_llm-train-*.parquet
- split: validation
path: data/primarily_llm-validation-*.parquet
- config_name: progressive
data_files:
- split: train
path: data/progressive-train-*.parquet
- split: validation
path: data/progressive-validation-*.parquet
- config_name: progressive_llm
data_files:
- split: train
path: data/progressive_llm-train-*.parquet
- split: validation
path: data/progressive_llm-validation-*.parquet
- config_name: prominence
data_files:
- split: train
path: data/prominence-train-*.parquet
- split: validation
path: data/prominence-validation-*.parquet
- config_name: prominence_llm
data_files:
- split: train
path: data/prominence_llm-train-*.parquet
- split: validation
path: data/prominence_llm-validation-*.parquet
- config_name: prose_chain_8
data_files:
- split: train
path: data/prose_chain_8-train-*.parquet
- split: validation
path: data/prose_chain_8-validation-*.parquet
- config_name: prose_chain_8_llm
data_files:
- split: train
path: data/prose_chain_8_llm-train-*.parquet
- split: validation
path: data/prose_chain_8_llm-validation-*.parquet
- config_name: prose_mid
data_files:
- split: train
path: data/prose_mid-train-*.parquet
- split: validation
path: data/prose_mid-validation-*.parquet
- config_name: prose_mid_llm
data_files:
- split: train
path: data/prose_mid_llm-train-*.parquet
- split: validation
path: data/prose_mid_llm-validation-*.parquet
- config_name: reasoning
data_files:
- split: train
path: data/reasoning-train-*.parquet
- split: validation
path: data/reasoning-validation-*.parquet
- config_name: reasoning_llm
data_files:
- split: train
path: data/reasoning_llm-train-*.parquet
- split: validation
path: data/reasoning_llm-validation-*.parquet
- config_name: redirect
data_files:
- split: train
path: data/redirect-train-*.parquet
- split: validation
path: data/redirect-validation-*.parquet
- config_name: redirect_llm
data_files:
- split: train
path: data/redirect_llm-train-*.parquet
- split: validation
path: data/redirect_llm-validation-*.parquet
- config_name: relative
data_files:
- split: train
path: data/relative-train-*.parquet
- split: validation
path: data/relative-validation-*.parquet
- config_name: relative_llm
data_files:
- split: train
path: data/relative_llm-train-*.parquet
- split: validation
path: data/relative_llm-validation-*.parquet
- config_name: rewrite_protect
data_files:
- split: train
path: data/rewrite_protect-train-*.parquet
- split: validation
path: data/rewrite_protect-validation-*.parquet
- config_name: rhetorical
data_files:
- split: train
path: data/rhetorical-train-*.parquet
- split: validation
path: data/rhetorical-validation-*.parquet
- config_name: rhetorical_llm
data_files:
- split: train
path: data/rhetorical_llm-train-*.parquet
- split: validation
path: data/rhetorical_llm-validation-*.parquet
- config_name: seq2sick
data_files:
- split: train
path: data/seq2sick-train-*.parquet
- split: validation
path: data/seq2sick-validation-*.parquet
- config_name: seq2sick_llm
data_files:
- split: train
path: data/seq2sick_llm-train-*.parquet
- split: validation
path: data/seq2sick_llm-validation-*.parquet
- config_name: seven_center
data_files:
- split: train
path: data/seven_center-train-*.parquet
- split: validation
path: data/seven_center-validation-*.parquet
- config_name: seven_center_llm
data_files:
- split: train
path: data/seven_center_llm-train-*.parquet
- split: validation
path: data/seven_center_llm-validation-*.parquet
- config_name: shadow
data_files:
- split: train
path: data/shadow-train-*.parquet
- split: validation
path: data/shadow-validation-*.parquet
- config_name: shadow_list
data_files:
- split: train
path: data/shadow_list-train-*.parquet
- split: validation
path: data/shadow_list-validation-*.parquet
- config_name: shadow_list_llm
data_files:
- split: train
path: data/shadow_list_llm-train-*.parquet
- split: validation
path: data/shadow_list_llm-validation-*.parquet
- config_name: shadow_llm
data_files:
- split: train
path: data/shadow_llm-train-*.parquet
- split: validation
path: data/shadow_llm-validation-*.parquet
- config_name: shortcorr
data_files:
- split: train
path: data/shortcorr-train-*.parquet
- split: validation
path: data/shortcorr-validation-*.parquet
- config_name: shortcorr_llm
data_files:
- split: train
path: data/shortcorr_llm-train-*.parquet
- split: validation
path: data/shortcorr_llm-validation-*.parquet
- config_name: subordinate
data_files:
- split: train
path: data/subordinate-train-*.parquet
- split: validation
path: data/subordinate-validation-*.parquet
- config_name: subordinate_llm
data_files:
- split: train
path: data/subordinate_llm-train-*.parquet
- split: validation
path: data/subordinate_llm-validation-*.parquet
- config_name: tail_restate
data_files:
- split: train
path: data/tail_restate-train-*.parquet
- split: validation
path: data/tail_restate-validation-*.parquet
- config_name: tail_restate_llm
data_files:
- split: train
path: data/tail_restate_llm-train-*.parquet
- split: validation
path: data/tail_restate_llm-validation-*.parquet
- config_name: tail_weighted_10
data_files:
- split: train
path: data/tail_weighted_10-train-*.parquet
- split: validation
path: data/tail_weighted_10-validation-*.parquet
- config_name: tail_weighted_10_llm
data_files:
- split: train
path: data/tail_weighted_10_llm-train-*.parquet
- split: validation
path: data/tail_weighted_10_llm-validation-*.parquet
- config_name: temporal
data_files:
- split: train
path: data/temporal-train-*.parquet
- split: validation
path: data/temporal-validation-*.parquet
- config_name: temporal_llm
data_files:
- split: train
path: data/temporal_llm-train-*.parquet
- split: validation
path: data/temporal_llm-validation-*.parquet
- config_name: ten_item_mid
data_files:
- split: train
path: data/ten_item_mid-train-*.parquet
- split: validation
path: data/ten_item_mid-validation-*.parquet
- config_name: ten_item_mid_llm
data_files:
- split: train
path: data/ten_item_mid_llm-train-*.parquet
- split: validation
path: data/ten_item_mid_llm-validation-*.parquet
- config_name: think
data_files:
- split: train
path: data/think-train-*.parquet
- split: validation
path: data/think-validation-*.parquet
- config_name: think_llm
data_files:
- split: train
path: data/think_llm-train-*.parquet
- split: validation
path: data/think_llm-validation-*.parquet
- config_name: trap
data_files:
- split: train
path: data/trap-train-*.parquet
- split: validation
path: data/trap-validation-*.parquet
- config_name: trap_llm
data_files:
- split: train
path: data/trap_llm-train-*.parquet
- split: validation
path: data/trap_llm-validation-*.parquet
- config_name: triple_decoy
data_files:
- split: train
path: data/triple_decoy-train-*.parquet
- split: validation
path: data/triple_decoy-validation-*.parquet
- config_name: triple_decoy_llm
data_files:
- split: train
path: data/triple_decoy_llm-train-*.parquet
- split: validation
path: data/triple_decoy_llm-validation-*.parquet
- config_name: triple_mid
data_files:
- split: train
path: data/triple_mid-train-*.parquet
- split: validation
path: data/triple_mid-validation-*.parquet
- config_name: triple_mid_llm
data_files:
- split: train
path: data/triple_mid_llm-train-*.parquet
- split: validation
path: data/triple_mid_llm-validation-*.parquet
- config_name: trivia
data_files:
- split: train
path: data/trivia-train-*.parquet
- split: validation
path: data/trivia-validation-*.parquet
- config_name: trivia_llm
data_files:
- split: train
path: data/trivia_llm-train-*.parquet
- split: validation
path: data/trivia_llm-validation-*.parquet
- config_name: unlabeled_10_mid
data_files:
- split: train
path: data/unlabeled_10_mid-train-*.parquet
- split: validation
path: data/unlabeled_10_mid-validation-*.parquet
- config_name: unlabeled_10_mid_llm
data_files:
- split: train
path: data/unlabeled_10_mid_llm-train-*.parquet
- split: validation
path: data/unlabeled_10_mid_llm-validation-*.parquet
- config_name: unlabeled_5
data_files:
- split: train
path: data/unlabeled_5-train-*.parquet
- split: validation
path: data/unlabeled_5-validation-*.parquet
- config_name: unlabeled_5_llm
data_files:
- split: train
path: data/unlabeled_5_llm-train-*.parquet
- split: validation
path: data/unlabeled_5_llm-validation-*.parquet
- config_name: unlabeled_7
data_files:
- split: train
path: data/unlabeled_7-train-*.parquet
- split: validation
path: data/unlabeled_7-validation-*.parquet
- config_name: unlabeled_7_llm
data_files:
- split: train
path: data/unlabeled_7_llm-train-*.parquet
- split: validation
path: data/unlabeled_7_llm-validation-*.parquet
- config_name: unlocks
data_files:
- split: train
path: data/unlocks-train-*.parquet
- split: validation
path: data/unlocks-validation-*.parquet
- config_name: unlocks_llm
data_files:
- split: train
path: data/unlocks_llm-train-*.parquet
- split: validation
path: data/unlocks_llm-validation-*.parquet
- config_name: verbose
data_files:
- split: train
path: data/verbose-train-*.parquet
- split: validation
path: data/verbose-validation-*.parquet
- config_name: verbose_llm
data_files:
- split: train
path: data/verbose_llm-train-*.parquet
- split: validation
path: data/verbose_llm-validation-*.parquet
- config_name: vs
data_files:
- split: train
path: data/vs-train-*.parquet
- split: validation
path: data/vs-validation-*.parquet
- config_name: vs_llm
data_files:
- split: train
path: data/vs_llm-train-*.parquet
- split: validation
path: data/vs_llm-validation-*.parquet
- config_name: whereas
data_files:
- split: train
path: data/whereas-train-*.parquet
- split: validation
path: data/whereas-validation-*.parquet
- config_name: whereas_llm
data_files:
- split: train
path: data/whereas_llm-train-*.parquet
- split: validation
path: data/whereas_llm-validation-*.parquet
- config_name: wiki
data_files:
- split: train
path: data/wiki-train-*.parquet
- split: validation
path: data/wiki-validation-*.parquet
- config_name: wiki_llm
data_files:
- split: train
path: data/wiki_llm-train-*.parquet
- split: validation
path: data/wiki_llm-validation-*.parquet
- config_name: wrap
data_files:
- split: train
path: data/wrap-train-*.parquet
- split: validation
path: data/wrap-validation-*.parquet
- config_name: wrap_llm
data_files:
- split: train
path: data/wrap_llm-train-*.parquet
- split: validation
path: data/wrap_llm-validation-*.parquet
---

# Variant Dataset

Flattened schema: each config has a single `answer` column.

| Column | Description |
|--------|-------------|
| `question` | the question |
| `answer` | variant text; always clean in the validation split |
| `prediction_lora` | LoRA model trained on this answer ("" if unavailable) |
| `prediction_freeze` | full SFT model trained on this answer |