wisent 0.7.701__py3-none-any.whl → 0.7.901__py3-none-any.whl
This diff represents the content of publicly available package versions that have been released to one of the supported registries. The information contained in this diff is provided for informational purposes only and reflects changes between package versions as they appear in their respective public registries.
- wisent/__init__.py +1 -1
- wisent/core/activations/activation_cache.py +393 -0
- wisent/core/activations/activations.py +3 -3
- wisent/core/activations/activations_collector.py +9 -5
- wisent/core/activations/classifier_inference_strategy.py +12 -11
- wisent/core/activations/extraction_strategy.py +256 -84
- wisent/core/classifiers/classifiers/core/atoms.py +3 -2
- wisent/core/cli/__init__.py +2 -1
- wisent/core/cli/agent/apply_steering.py +5 -7
- wisent/core/cli/agent/train_classifier.py +19 -7
- wisent/core/cli/check_linearity.py +35 -3
- wisent/core/cli/cluster_benchmarks.py +4 -6
- wisent/core/cli/create_steering_vector.py +6 -4
- wisent/core/cli/diagnose_vectors.py +7 -4
- wisent/core/cli/estimate_unified_goodness_time.py +6 -4
- wisent/core/cli/generate_pairs_from_task.py +9 -56
- wisent/core/cli/geometry_search.py +137 -0
- wisent/core/cli/get_activations.py +1 -1
- wisent/core/cli/method_optimizer.py +4 -3
- wisent/core/cli/modify_weights.py +3 -2
- wisent/core/cli/optimize_sample_size.py +1 -1
- wisent/core/cli/optimize_steering.py +14 -16
- wisent/core/cli/optimize_weights.py +2 -1
- wisent/core/cli/preview_pairs.py +203 -0
- wisent/core/cli/steering_method_trainer.py +3 -3
- wisent/core/cli/tasks.py +19 -76
- wisent/core/cli/train_unified_goodness.py +3 -3
- wisent/core/contrastive_pairs/diagnostics/control_vectors.py +4 -4
- wisent/core/contrastive_pairs/diagnostics/linearity.py +7 -0
- wisent/core/contrastive_pairs/huggingface_pairs/hf_task_extractors/agentic_search.py +37 -347
- wisent/core/contrastive_pairs/huggingface_pairs/hf_task_extractors/aider_polyglot.py +113 -136
- wisent/core/contrastive_pairs/huggingface_pairs/hf_task_extractors/codeforces.py +2 -12
- wisent/core/contrastive_pairs/huggingface_pairs/hf_task_extractors/coding_benchmarks.py +124 -504
- wisent/core/contrastive_pairs/huggingface_pairs/hf_task_extractors/faithbench.py +40 -63
- wisent/core/contrastive_pairs/huggingface_pairs/hf_task_extractors/flames.py +46 -89
- wisent/core/contrastive_pairs/huggingface_pairs/hf_task_extractors/flores.py +15 -4
- wisent/core/contrastive_pairs/huggingface_pairs/hf_task_extractors/frames.py +36 -20
- wisent/core/contrastive_pairs/huggingface_pairs/hf_task_extractors/hallucinations_leaderboard.py +3 -45
- wisent/core/contrastive_pairs/huggingface_pairs/hf_task_extractors/livemathbench.py +42 -4
- wisent/core/contrastive_pairs/huggingface_pairs/hf_task_extractors/longform_writing.py +2 -112
- wisent/core/contrastive_pairs/huggingface_pairs/hf_task_extractors/math500.py +39 -4
- wisent/core/contrastive_pairs/huggingface_pairs/hf_task_extractors/medium_priority_benchmarks.py +475 -525
- wisent/core/contrastive_pairs/huggingface_pairs/hf_task_extractors/mercury.py +65 -42
- wisent/core/contrastive_pairs/huggingface_pairs/hf_task_extractors/olympiadbench.py +2 -12
- wisent/core/contrastive_pairs/huggingface_pairs/hf_task_extractors/planbench.py +78 -219
- wisent/core/contrastive_pairs/huggingface_pairs/hf_task_extractors/polymath.py +37 -4
- wisent/core/contrastive_pairs/huggingface_pairs/hf_task_extractors/recode.py +84 -69
- wisent/core/contrastive_pairs/huggingface_pairs/hf_task_extractors/refusalbench.py +168 -160
- wisent/core/contrastive_pairs/huggingface_pairs/hf_task_extractors/simpleqa.py +44 -25
- wisent/core/contrastive_pairs/huggingface_pairs/hf_task_extractors/tau_bench.py +3 -103
- wisent/core/contrastive_pairs/huggingface_pairs/hf_task_extractors/toolbench.py +3 -97
- wisent/core/contrastive_pairs/huggingface_pairs/hf_task_extractors/toolemu.py +48 -182
- wisent/core/contrastive_pairs/lm_eval_pairs/lm_extractor_manifest.py +3 -0
- wisent/core/contrastive_pairs/lm_eval_pairs/lm_extractor_registry.py +19 -1
- wisent/core/contrastive_pairs/lm_eval_pairs/lm_task_extractors/aclue.py +1 -3
- wisent/core/contrastive_pairs/lm_eval_pairs/lm_task_extractors/acp_bench.py +1 -3
- wisent/core/contrastive_pairs/lm_eval_pairs/lm_task_extractors/acp_bench_hard.py +1 -3
- wisent/core/contrastive_pairs/lm_eval_pairs/lm_task_extractors/advanced.py +2 -4
- wisent/core/contrastive_pairs/lm_eval_pairs/lm_task_extractors/aexams.py +1 -3
- wisent/core/contrastive_pairs/lm_eval_pairs/lm_task_extractors/afrimmlu.py +1 -3
- wisent/core/contrastive_pairs/lm_eval_pairs/lm_task_extractors/afrixnli.py +2 -2
- wisent/core/contrastive_pairs/lm_eval_pairs/lm_task_extractors/arabculture.py +1 -3
- wisent/core/contrastive_pairs/lm_eval_pairs/lm_task_extractors/arabic.py +1 -3
- wisent/core/contrastive_pairs/lm_eval_pairs/lm_task_extractors/arabic_exams.py +1 -3
- wisent/core/contrastive_pairs/lm_eval_pairs/lm_task_extractors/arabic_leaderboard_complete.py +1 -3
- wisent/core/contrastive_pairs/lm_eval_pairs/lm_task_extractors/arabic_leaderboard_light.py +1 -3
- wisent/core/contrastive_pairs/lm_eval_pairs/lm_task_extractors/arabicmmlu.py +1 -3
- wisent/core/contrastive_pairs/lm_eval_pairs/lm_task_extractors/aradice.py +1 -3
- wisent/core/contrastive_pairs/lm_eval_pairs/lm_task_extractors/arc.py +1 -3
- wisent/core/contrastive_pairs/lm_eval_pairs/lm_task_extractors/arc_challenge.py +1 -2
- wisent/core/contrastive_pairs/lm_eval_pairs/lm_task_extractors/arc_easy.py +1 -2
- wisent/core/contrastive_pairs/lm_eval_pairs/lm_task_extractors/arithmetic.py +2 -2
- wisent/core/contrastive_pairs/lm_eval_pairs/lm_task_extractors/asdiv.py +2 -2
- wisent/core/contrastive_pairs/lm_eval_pairs/lm_task_extractors/babi.py +36 -2
- wisent/core/contrastive_pairs/lm_eval_pairs/lm_task_extractors/basque_bench.py +1 -3
- wisent/core/contrastive_pairs/lm_eval_pairs/lm_task_extractors/bbq.py +1 -3
- wisent/core/contrastive_pairs/lm_eval_pairs/lm_task_extractors/belebele.py +1 -3
- wisent/core/contrastive_pairs/lm_eval_pairs/lm_task_extractors/benchmarks.py +1 -3
- wisent/core/contrastive_pairs/lm_eval_pairs/lm_task_extractors/bertaqa.py +1 -3
- wisent/core/contrastive_pairs/lm_eval_pairs/lm_task_extractors/bhs.py +1 -3
- wisent/core/contrastive_pairs/lm_eval_pairs/lm_task_extractors/bhtc.py +3 -5
- wisent/core/contrastive_pairs/lm_eval_pairs/lm_task_extractors/blimp.py +1 -3
- wisent/core/contrastive_pairs/lm_eval_pairs/lm_task_extractors/blimp_nl.py +1 -3
- wisent/core/contrastive_pairs/lm_eval_pairs/lm_task_extractors/boolq.py +2 -2
- wisent/core/contrastive_pairs/lm_eval_pairs/lm_task_extractors/c4.py +1 -3
- wisent/core/contrastive_pairs/lm_eval_pairs/lm_task_extractors/cabbq.py +1 -3
- wisent/core/contrastive_pairs/lm_eval_pairs/lm_task_extractors/careqa.py +1 -3
- wisent/core/contrastive_pairs/lm_eval_pairs/lm_task_extractors/catalan_bench.py +1 -3
- wisent/core/contrastive_pairs/lm_eval_pairs/lm_task_extractors/catalanqa.py +1 -3
- wisent/core/contrastive_pairs/lm_eval_pairs/lm_task_extractors/catcola.py +1 -3
- wisent/core/contrastive_pairs/lm_eval_pairs/lm_task_extractors/cb.py +2 -2
- wisent/core/contrastive_pairs/lm_eval_pairs/lm_task_extractors/ceval.py +1 -3
- wisent/core/contrastive_pairs/lm_eval_pairs/lm_task_extractors/ceval_valid.py +1 -3
- wisent/core/contrastive_pairs/lm_eval_pairs/lm_task_extractors/chain.py +1 -3
- wisent/core/contrastive_pairs/lm_eval_pairs/lm_task_extractors/chartqa.py +1 -3
- wisent/core/contrastive_pairs/lm_eval_pairs/lm_task_extractors/claim.py +1 -3
- wisent/core/contrastive_pairs/lm_eval_pairs/lm_task_extractors/click.py +1 -3
- wisent/core/contrastive_pairs/lm_eval_pairs/lm_task_extractors/cmmlu.py +1 -3
- wisent/core/contrastive_pairs/lm_eval_pairs/lm_task_extractors/cnn.py +1 -3
- wisent/core/contrastive_pairs/lm_eval_pairs/lm_task_extractors/cocoteros.py +1 -3
- wisent/core/contrastive_pairs/lm_eval_pairs/lm_task_extractors/coedit.py +1 -3
- wisent/core/contrastive_pairs/lm_eval_pairs/lm_task_extractors/commonsense.py +1 -3
- wisent/core/contrastive_pairs/lm_eval_pairs/lm_task_extractors/commonsense_qa.py +1 -3
- wisent/core/contrastive_pairs/lm_eval_pairs/lm_task_extractors/copa.py +2 -2
- wisent/core/contrastive_pairs/lm_eval_pairs/lm_task_extractors/copal_id.py +1 -3
- wisent/core/contrastive_pairs/lm_eval_pairs/lm_task_extractors/coqa.py +3 -4
- wisent/core/contrastive_pairs/lm_eval_pairs/lm_task_extractors/csatqa.py +1 -3
- wisent/core/contrastive_pairs/lm_eval_pairs/lm_task_extractors/cycle.py +1 -3
- wisent/core/contrastive_pairs/lm_eval_pairs/lm_task_extractors/darija_bench.py +1 -3
- wisent/core/contrastive_pairs/lm_eval_pairs/lm_task_extractors/darijahellaswag.py +2 -6
- wisent/core/contrastive_pairs/lm_eval_pairs/lm_task_extractors/darijammlu.py +1 -3
- wisent/core/contrastive_pairs/lm_eval_pairs/lm_task_extractors/dbpedia.py +1 -3
- wisent/core/contrastive_pairs/lm_eval_pairs/lm_task_extractors/discrim_eval.py +1 -3
- wisent/core/contrastive_pairs/lm_eval_pairs/lm_task_extractors/doc.py +1 -3
- wisent/core/contrastive_pairs/lm_eval_pairs/lm_task_extractors/drop.py +2 -2
- wisent/core/contrastive_pairs/lm_eval_pairs/lm_task_extractors/epec.py +1 -3
- wisent/core/contrastive_pairs/lm_eval_pairs/lm_task_extractors/eq.py +1 -3
- wisent/core/contrastive_pairs/lm_eval_pairs/lm_task_extractors/eq_bench.py +1 -3
- wisent/core/contrastive_pairs/lm_eval_pairs/lm_task_extractors/eq_bench_ca.py +1 -3
- wisent/core/contrastive_pairs/lm_eval_pairs/lm_task_extractors/eq_bench_es.py +1 -3
- wisent/core/contrastive_pairs/lm_eval_pairs/lm_task_extractors/esbbq.py +1 -3
- wisent/core/contrastive_pairs/lm_eval_pairs/lm_task_extractors/ethics.py +1 -3
- wisent/core/contrastive_pairs/lm_eval_pairs/lm_task_extractors/eus.py +1 -3
- wisent/core/contrastive_pairs/lm_eval_pairs/lm_task_extractors/eus_exams.py +1 -3
- wisent/core/contrastive_pairs/lm_eval_pairs/lm_task_extractors/eus_proficiency.py +1 -3
- wisent/core/contrastive_pairs/lm_eval_pairs/lm_task_extractors/eus_reading.py +1 -3
- wisent/core/contrastive_pairs/lm_eval_pairs/lm_task_extractors/eus_trivia.py +1 -3
- wisent/core/contrastive_pairs/lm_eval_pairs/lm_task_extractors/evalita_llm.py +1 -3
- wisent/core/contrastive_pairs/lm_eval_pairs/lm_task_extractors/financial.py +1 -3
- wisent/core/contrastive_pairs/lm_eval_pairs/lm_task_extractors/flan.py +1 -3
- wisent/core/contrastive_pairs/lm_eval_pairs/lm_task_extractors/french_bench.py +1 -3
- wisent/core/contrastive_pairs/lm_eval_pairs/lm_task_extractors/galician_bench.py +1 -3
- wisent/core/contrastive_pairs/lm_eval_pairs/lm_task_extractors/gaokao.py +2 -2
- wisent/core/contrastive_pairs/lm_eval_pairs/lm_task_extractors/glianorex.py +1 -3
- wisent/core/contrastive_pairs/lm_eval_pairs/lm_task_extractors/global_mmlu.py +1 -3
- wisent/core/contrastive_pairs/lm_eval_pairs/lm_task_extractors/global_piqa.py +1 -3
- wisent/core/contrastive_pairs/lm_eval_pairs/lm_task_extractors/gpt3.py +1 -3
- wisent/core/contrastive_pairs/lm_eval_pairs/lm_task_extractors/groundcocoa.py +1 -3
- wisent/core/contrastive_pairs/lm_eval_pairs/lm_task_extractors/haerae.py +1 -3
- wisent/core/contrastive_pairs/lm_eval_pairs/lm_task_extractors/headqa.py +2 -2
- wisent/core/contrastive_pairs/lm_eval_pairs/lm_task_extractors/hellaswag.py +2 -2
- wisent/core/contrastive_pairs/lm_eval_pairs/lm_task_extractors/hendrycks_ethics.py +5 -9
- wisent/core/contrastive_pairs/lm_eval_pairs/lm_task_extractors/hendrycks_math.py +63 -16
- wisent/core/contrastive_pairs/lm_eval_pairs/lm_task_extractors/histoires_morales.py +1 -3
- wisent/core/contrastive_pairs/lm_eval_pairs/lm_task_extractors/hrm8k.py +1 -3
- wisent/core/contrastive_pairs/lm_eval_pairs/lm_task_extractors/humaneval_infilling.py +1 -3
- wisent/core/contrastive_pairs/lm_eval_pairs/lm_task_extractors/icelandic_winogrande.py +1 -3
- wisent/core/contrastive_pairs/lm_eval_pairs/lm_task_extractors/inverse.py +1 -3
- wisent/core/contrastive_pairs/lm_eval_pairs/lm_task_extractors/inverse_scaling.py +1 -3
- wisent/core/contrastive_pairs/lm_eval_pairs/lm_task_extractors/ja.py +1 -3
- wisent/core/contrastive_pairs/lm_eval_pairs/lm_task_extractors/japanese_leaderboard.py +1 -3
- wisent/core/contrastive_pairs/lm_eval_pairs/lm_task_extractors/japanese_leaderboard_mc.py +1 -1
- wisent/core/contrastive_pairs/lm_eval_pairs/lm_task_extractors/kmmlu.py +1 -3
- wisent/core/contrastive_pairs/lm_eval_pairs/lm_task_extractors/kobest.py +1 -3
- wisent/core/contrastive_pairs/lm_eval_pairs/lm_task_extractors/kormedmcqa.py +5 -17
- wisent/core/contrastive_pairs/lm_eval_pairs/lm_task_extractors/lambada_cloze.py +1 -3
- wisent/core/contrastive_pairs/lm_eval_pairs/lm_task_extractors/lambada_multilingual.py +1 -3
- wisent/core/contrastive_pairs/lm_eval_pairs/lm_task_extractors/law.py +1 -3
- wisent/core/contrastive_pairs/lm_eval_pairs/lm_task_extractors/leaderboard.py +1 -3
- wisent/core/contrastive_pairs/lm_eval_pairs/lm_task_extractors/lingoly.py +1 -3
- wisent/core/contrastive_pairs/lm_eval_pairs/lm_task_extractors/llama3.py +1 -3
- wisent/core/contrastive_pairs/lm_eval_pairs/lm_task_extractors/lm_syneval.py +1 -3
- wisent/core/contrastive_pairs/lm_eval_pairs/lm_task_extractors/logiqa.py +2 -2
- wisent/core/contrastive_pairs/lm_eval_pairs/lm_task_extractors/logiqa2.py +2 -2
- wisent/core/contrastive_pairs/lm_eval_pairs/lm_task_extractors/longbench.py +1 -3
- wisent/core/contrastive_pairs/lm_eval_pairs/lm_task_extractors/longbenchv2.py +1 -3
- wisent/core/contrastive_pairs/lm_eval_pairs/lm_task_extractors/mastermind.py +2 -4
- wisent/core/contrastive_pairs/lm_eval_pairs/lm_task_extractors/mc-taco.py +2 -2
- wisent/core/contrastive_pairs/lm_eval_pairs/lm_task_extractors/med_concepts_qa.py +2 -4
- wisent/core/contrastive_pairs/lm_eval_pairs/lm_task_extractors/meddialog.py +1 -3
- wisent/core/contrastive_pairs/lm_eval_pairs/lm_task_extractors/medical.py +1 -3
- wisent/core/contrastive_pairs/lm_eval_pairs/lm_task_extractors/medmcqa.py +1 -3
- wisent/core/contrastive_pairs/lm_eval_pairs/lm_task_extractors/medqa.py +2 -2
- wisent/core/contrastive_pairs/lm_eval_pairs/lm_task_extractors/mela.py +2 -2
- wisent/core/contrastive_pairs/lm_eval_pairs/lm_task_extractors/metabench.py +1 -3
- wisent/core/contrastive_pairs/lm_eval_pairs/lm_task_extractors/minerva_math.py +1 -3
- wisent/core/contrastive_pairs/lm_eval_pairs/lm_task_extractors/mmlu.py +1 -3
- wisent/core/contrastive_pairs/lm_eval_pairs/lm_task_extractors/mmlusr.py +3 -4
- wisent/core/contrastive_pairs/lm_eval_pairs/lm_task_extractors/mrpc.py +2 -2
- wisent/core/contrastive_pairs/lm_eval_pairs/lm_task_extractors/multiblimp.py +2 -5
- wisent/core/contrastive_pairs/lm_eval_pairs/lm_task_extractors/multirc.py +2 -2
- wisent/core/contrastive_pairs/lm_eval_pairs/lm_task_extractors/mutual.py +2 -2
- wisent/core/contrastive_pairs/lm_eval_pairs/lm_task_extractors/non.py +1 -3
- wisent/core/contrastive_pairs/lm_eval_pairs/lm_task_extractors/noreval.py +1 -3
- wisent/core/contrastive_pairs/lm_eval_pairs/lm_task_extractors/noreval_exact.py +1 -3
- wisent/core/contrastive_pairs/lm_eval_pairs/lm_task_extractors/noreval_gen_exact.py +1 -3
- wisent/core/contrastive_pairs/lm_eval_pairs/lm_task_extractors/noreval_mc.py +4 -8
- wisent/core/contrastive_pairs/lm_eval_pairs/lm_task_extractors/noreval_mc_log_likelihoods.py +4 -8
- wisent/core/contrastive_pairs/lm_eval_pairs/lm_task_extractors/nq_open.py +2 -2
- wisent/core/contrastive_pairs/lm_eval_pairs/lm_task_extractors/okapi_arc_multilingual.py +1 -3
- wisent/core/contrastive_pairs/lm_eval_pairs/lm_task_extractors/okapi_hellaswag_multilingual.py +1 -3
- wisent/core/contrastive_pairs/lm_eval_pairs/lm_task_extractors/okapi_mmlu_multilingual.py +1 -3
- wisent/core/contrastive_pairs/lm_eval_pairs/lm_task_extractors/okapi_truthfulqa_multilingual.py +2 -5
- wisent/core/contrastive_pairs/lm_eval_pairs/lm_task_extractors/olaph.py +1 -3
- wisent/core/contrastive_pairs/lm_eval_pairs/lm_task_extractors/openbookqa.py +2 -2
- wisent/core/contrastive_pairs/lm_eval_pairs/lm_task_extractors/option.py +1 -3
- wisent/core/contrastive_pairs/lm_eval_pairs/lm_task_extractors/parafraseja.py +1 -3
- wisent/core/contrastive_pairs/lm_eval_pairs/lm_task_extractors/parafrases.py +1 -3
- wisent/core/contrastive_pairs/lm_eval_pairs/lm_task_extractors/paws.py +1 -3
- wisent/core/contrastive_pairs/lm_eval_pairs/lm_task_extractors/paws_x.py +1 -3
- wisent/core/contrastive_pairs/lm_eval_pairs/lm_task_extractors/pawsx.py +2 -2
- wisent/core/contrastive_pairs/lm_eval_pairs/lm_task_extractors/persona.py +1 -3
- wisent/core/contrastive_pairs/lm_eval_pairs/lm_task_extractors/phrases.py +1 -3
- wisent/core/contrastive_pairs/lm_eval_pairs/lm_task_extractors/pile.py +1 -3
- wisent/core/contrastive_pairs/lm_eval_pairs/lm_task_extractors/piqa.py +2 -2
- wisent/core/contrastive_pairs/lm_eval_pairs/lm_task_extractors/portuguese_bench.py +1 -3
- wisent/core/contrastive_pairs/lm_eval_pairs/lm_task_extractors/prompt.py +1 -3
- wisent/core/contrastive_pairs/lm_eval_pairs/lm_task_extractors/prost.py +2 -2
- wisent/core/contrastive_pairs/lm_eval_pairs/lm_task_extractors/pubmedqa.py +2 -2
- wisent/core/contrastive_pairs/lm_eval_pairs/lm_task_extractors/qa4mre.py +2 -2
- wisent/core/contrastive_pairs/lm_eval_pairs/lm_task_extractors/qasper.py +2 -2
- wisent/core/contrastive_pairs/lm_eval_pairs/lm_task_extractors/qasper_bool.py +2 -2
- wisent/core/contrastive_pairs/lm_eval_pairs/lm_task_extractors/qnli.py +2 -2
- wisent/core/contrastive_pairs/lm_eval_pairs/lm_task_extractors/qnlieu.py +1 -3
- wisent/core/contrastive_pairs/lm_eval_pairs/lm_task_extractors/qqp.py +2 -2
- wisent/core/contrastive_pairs/lm_eval_pairs/lm_task_extractors/race.py +2 -2
- wisent/core/contrastive_pairs/lm_eval_pairs/lm_task_extractors/random.py +1 -3
- wisent/core/contrastive_pairs/lm_eval_pairs/lm_task_extractors/record.py +2 -2
- wisent/core/contrastive_pairs/lm_eval_pairs/lm_task_extractors/reversed.py +1 -3
- wisent/core/contrastive_pairs/lm_eval_pairs/lm_task_extractors/rte.py +2 -2
- wisent/core/contrastive_pairs/lm_eval_pairs/lm_task_extractors/ruler.py +1 -3
- wisent/core/contrastive_pairs/lm_eval_pairs/lm_task_extractors/sciq.py +2 -2
- wisent/core/contrastive_pairs/lm_eval_pairs/lm_task_extractors/score.py +1 -3
- wisent/core/contrastive_pairs/lm_eval_pairs/lm_task_extractors/scrolls.py +1 -3
- wisent/core/contrastive_pairs/lm_eval_pairs/lm_task_extractors/scrolls_mc.py +1 -3
- wisent/core/contrastive_pairs/lm_eval_pairs/lm_task_extractors/self.py +1 -3
- wisent/core/contrastive_pairs/lm_eval_pairs/lm_task_extractors/sglue.py +1 -3
- wisent/core/contrastive_pairs/lm_eval_pairs/lm_task_extractors/sglue_rte.py +2 -1
- wisent/core/contrastive_pairs/lm_eval_pairs/lm_task_extractors/siqa.py +4 -7
- wisent/core/contrastive_pairs/lm_eval_pairs/lm_task_extractors/social_iqa.py +2 -2
- wisent/core/contrastive_pairs/lm_eval_pairs/lm_task_extractors/spanish_bench.py +1 -3
- wisent/core/contrastive_pairs/lm_eval_pairs/lm_task_extractors/storycloze.py +2 -6
- wisent/core/contrastive_pairs/lm_eval_pairs/lm_task_extractors/summarization.py +1 -3
- wisent/core/contrastive_pairs/lm_eval_pairs/lm_task_extractors/super.py +1 -3
- wisent/core/contrastive_pairs/lm_eval_pairs/lm_task_extractors/super_glue.py +1 -3
- wisent/core/contrastive_pairs/lm_eval_pairs/lm_task_extractors/swag.py +2 -2
- wisent/core/contrastive_pairs/lm_eval_pairs/lm_task_extractors/swde.py +1 -3
- wisent/core/contrastive_pairs/lm_eval_pairs/lm_task_extractors/sycophancy.py +1 -3
- wisent/core/contrastive_pairs/lm_eval_pairs/lm_task_extractors/t0.py +1 -3
- wisent/core/contrastive_pairs/lm_eval_pairs/lm_task_extractors/teca.py +1 -3
- wisent/core/contrastive_pairs/lm_eval_pairs/lm_task_extractors/tinyarc.py +1 -3
- wisent/core/contrastive_pairs/lm_eval_pairs/lm_task_extractors/tinybenchmarks.py +1 -3
- wisent/core/contrastive_pairs/lm_eval_pairs/lm_task_extractors/tinygsm8k.py +1 -3
- wisent/core/contrastive_pairs/lm_eval_pairs/lm_task_extractors/tinyhellaswag.py +1 -3
- wisent/core/contrastive_pairs/lm_eval_pairs/lm_task_extractors/tinymmlu.py +1 -3
- wisent/core/contrastive_pairs/lm_eval_pairs/lm_task_extractors/tinytruthfulqa.py +1 -3
- wisent/core/contrastive_pairs/lm_eval_pairs/lm_task_extractors/tinywinogrande.py +1 -3
- wisent/core/contrastive_pairs/lm_eval_pairs/lm_task_extractors/tmmluplus.py +1 -3
- wisent/core/contrastive_pairs/lm_eval_pairs/lm_task_extractors/triviaqa.py +2 -2
- wisent/core/contrastive_pairs/lm_eval_pairs/lm_task_extractors/truthfulqa.py +1 -3
- wisent/core/contrastive_pairs/lm_eval_pairs/lm_task_extractors/truthfulqa_mc1.py +1 -3
- wisent/core/contrastive_pairs/lm_eval_pairs/lm_task_extractors/truthfulqa_mc2.py +1 -3
- wisent/core/contrastive_pairs/lm_eval_pairs/lm_task_extractors/turblimp_core.py +1 -3
- wisent/core/contrastive_pairs/lm_eval_pairs/lm_task_extractors/turkishmmlu.py +1 -3
- wisent/core/contrastive_pairs/lm_eval_pairs/lm_task_extractors/turkishmmlu_mc.py +0 -2
- wisent/core/contrastive_pairs/lm_eval_pairs/lm_task_extractors/unscramble.py +1 -3
- wisent/core/contrastive_pairs/lm_eval_pairs/lm_task_extractors/vaxx.py +2 -2
- wisent/core/contrastive_pairs/lm_eval_pairs/lm_task_extractors/webqs.py +2 -2
- wisent/core/contrastive_pairs/lm_eval_pairs/lm_task_extractors/wic.py +3 -4
- wisent/core/contrastive_pairs/lm_eval_pairs/lm_task_extractors/winogrande.py +2 -2
- wisent/core/contrastive_pairs/lm_eval_pairs/lm_task_extractors/wmdp.py +1 -3
- wisent/core/contrastive_pairs/lm_eval_pairs/lm_task_extractors/wnli.py +2 -2
- wisent/core/contrastive_pairs/lm_eval_pairs/lm_task_extractors/wsc.py +2 -2
- wisent/core/contrastive_pairs/lm_eval_pairs/lm_task_extractors/wsc273.py +1 -3
- wisent/core/contrastive_pairs/lm_eval_pairs/lm_task_extractors/xcopa.py +1 -3
- wisent/core/contrastive_pairs/lm_eval_pairs/lm_task_extractors/xlsum.py +1 -3
- wisent/core/contrastive_pairs/lm_eval_pairs/lm_task_extractors/xnli.py +2 -2
- wisent/core/contrastive_pairs/lm_eval_pairs/lm_task_extractors/xquad.py +2 -4
- wisent/core/contrastive_pairs/lm_eval_pairs/lm_task_extractors/xstorycloze.py +2 -3
- wisent/core/contrastive_pairs/lm_eval_pairs/lm_task_extractors/xwinograd.py +2 -2
- wisent/core/contrastive_pairs/lm_eval_pairs/lm_task_extractors/zhoblimp.py +1 -3
- wisent/core/contrastive_pairs/lm_eval_pairs/lm_task_pairs_generation.py +173 -6
- wisent/core/data_loaders/loaders/lm_loader.py +12 -1
- wisent/core/geometry_runner.py +995 -0
- wisent/core/geometry_search_space.py +237 -0
- wisent/core/hyperparameter_optimizer.py +1 -1
- wisent/core/main.py +3 -0
- wisent/core/models/core/atoms.py +5 -3
- wisent/core/models/wisent_model.py +1 -1
- wisent/core/optuna/classifier/optuna_classifier_optimizer.py +2 -2
- wisent/core/parser_arguments/check_linearity_parser.py +12 -2
- wisent/core/parser_arguments/generate_vector_from_synthetic_parser.py +2 -2
- wisent/core/parser_arguments/generate_vector_from_task_parser.py +2 -2
- wisent/core/parser_arguments/geometry_search_parser.py +61 -0
- wisent/core/parser_arguments/main_parser.py +8 -0
- wisent/core/parser_arguments/train_unified_goodness_parser.py +2 -2
- wisent/core/steering.py +5 -3
- wisent/core/steering_methods/methods/hyperplane.py +2 -1
- wisent/core/synthetic/generators/nonsense_generator.py +30 -18
- wisent/core/trainers/steering_trainer.py +2 -2
- wisent/core/utils/device.py +27 -27
- wisent/core/utils/layer_combinations.py +70 -0
- wisent/examples/__init__.py +1 -0
- wisent/examples/scripts/__init__.py +1 -0
- wisent/examples/scripts/count_all_benchmarks.py +121 -0
- wisent/examples/scripts/discover_directions.py +469 -0
- wisent/examples/scripts/extract_benchmark_info.py +71 -0
- wisent/examples/scripts/generate_paper_data.py +384 -0
- wisent/examples/scripts/intervention_validation.py +626 -0
- wisent/examples/scripts/results/test_AraDiCE_ArabicMMLU_lev_evaluation.json +324 -0
- wisent/examples/scripts/results/test_AraDiCE_ArabicMMLU_lev_pairs.json +92 -0
- wisent/examples/scripts/results/test_aexams_IslamicStudies_evaluation.json +324 -0
- wisent/examples/scripts/results/test_aexams_IslamicStudies_pairs.json +92 -0
- wisent/examples/scripts/results/test_afrimgsm_pairs.json +92 -0
- wisent/examples/scripts/results/test_afrimmlu_evaluation.json +324 -0
- wisent/examples/scripts/results/test_afrimmlu_pairs.json +92 -0
- wisent/examples/scripts/search_all_short_names.py +31 -0
- wisent/examples/scripts/test_all_benchmarks.py +138 -0
- wisent/examples/scripts/test_all_benchmarks_new.py +28 -0
- wisent/examples/scripts/test_contrastive_pairs_all_supported.py +230 -0
- wisent/examples/scripts/test_nonsense_baseline.py +261 -0
- wisent/examples/scripts/test_one_benchmark.py +324 -0
- wisent/examples/scripts/test_one_coding_benchmark.py +293 -0
- wisent/examples/scripts/threshold_analysis.py +434 -0
- wisent/examples/scripts/visualization_gallery.py +582 -0
- wisent/parameters/lm_eval/broken_in_lm_eval.json +179 -2
- wisent/parameters/lm_eval/category_directions.json +137 -0
- wisent/parameters/lm_eval/repair_plan.json +282 -0
- wisent/parameters/lm_eval/weak_contrastive_pairs.json +38 -0
- wisent/parameters/lm_eval/working_benchmarks.json +206 -0
- wisent/parameters/lm_eval/working_benchmarks_categorized.json +236 -0
- wisent/tests/test_detector_accuracy.py +1 -1
- wisent/tests/visualize_geometry.py +1 -1
- {wisent-0.7.701.dist-info → wisent-0.7.901.dist-info}/METADATA +1 -1
- {wisent-0.7.701.dist-info → wisent-0.7.901.dist-info}/RECORD +329 -295
- wisent/core/contrastive_pairs/huggingface_pairs/hf_task_extractors/browsecomp.py +0 -245
- {wisent-0.7.701.dist-info → wisent-0.7.901.dist-info}/WHEEL +0 -0
- {wisent-0.7.701.dist-info → wisent-0.7.901.dist-info}/entry_points.txt +0 -0
- {wisent-0.7.701.dist-info → wisent-0.7.901.dist-info}/licenses/LICENSE +0 -0
- {wisent-0.7.701.dist-info → wisent-0.7.901.dist-info}/top_level.txt +0 -0
|
@@ -0,0 +1,324 @@
|
|
|
1
|
+
{
|
|
2
|
+
"task_name": "aexams_IslamicStudies",
|
|
3
|
+
"model_name": "mock",
|
|
4
|
+
"evaluator_name": "log_likelihoods",
|
|
5
|
+
"num_pairs": 15,
|
|
6
|
+
"all_correct": true,
|
|
7
|
+
"pairs": [
|
|
8
|
+
{
|
|
9
|
+
"pair_id": 0,
|
|
10
|
+
"prompt": "\u0645\u0646 \u0634\u0631\u0648\u0637 \u0635\u062d\u0629 \u0627\u0644\u0635\u0644\u0627\u0629 ......",
|
|
11
|
+
"positive_response": "\u0627\u0633\u062a\u0642\u0628\u0627\u0644 \u0627\u0644\u0642\u0628\u0644\u0629.",
|
|
12
|
+
"negative_response": "\u0642\u0648\u0644 \u0633\u0628\u062d\u0627\u0646 \u0631\u0628\u064a \u0627\u0644\u0639\u0638\u064a\u0645.",
|
|
13
|
+
"positive_evaluation": {
|
|
14
|
+
"ground_truth": "TRUTHFUL",
|
|
15
|
+
"confidence": 1.0,
|
|
16
|
+
"expected": "TRUTHFUL",
|
|
17
|
+
"correct": true,
|
|
18
|
+
"details": "Predicted: '\u0627\u0633\u062a\u0642\u0628\u0627\u0644 \u0627\u0644\u0642\u0628\u0644\u0629.' (log_prob=-0.500), Expected: '\u0627\u0633\u062a\u0642\u0628\u0627\u0644 \u0627\u0644\u0642\u0628\u0644\u0629.'"
|
|
19
|
+
},
|
|
20
|
+
"negative_evaluation": {
|
|
21
|
+
"ground_truth": "UNTRUTHFUL",
|
|
22
|
+
"confidence": 0.0,
|
|
23
|
+
"expected": "UNTRUTHFUL",
|
|
24
|
+
"correct": true,
|
|
25
|
+
"details": "Predicted: '\u0627\u0633\u062a\u0642\u0628\u0627\u0644 \u0627\u0644\u0642\u0628\u0644\u0629.' (log_prob=-0.500), Expected: '\u0642\u0648\u0644 \u0633\u0628\u062d\u0627\u0646 \u0631\u0628\u064a \u0627\u0644\u0639\u0638\u064a\u0645.'"
|
|
26
|
+
},
|
|
27
|
+
"both_correct": true
|
|
28
|
+
},
|
|
29
|
+
{
|
|
30
|
+
"pair_id": 1,
|
|
31
|
+
"prompt": "\u0645\u0646 \u0627\u0644\u0623\u0633\u0628\u0627\u0628 \u0627\u0644\u062a\u064a \u062a\u0642\u0648\u064a \u0627\u0644\u0645\u0624\u0645\u0646 \u0641\u064a \u0633\u0644\u0648\u0643\u0647 :...",
|
|
32
|
+
"positive_response": "\u0627\u0644\u062a\u0641\u0642\u0647 \u0641\u064a \u0627\u0644\u062f\u064a\u0646.",
|
|
33
|
+
"negative_response": "\u0645\u0645\u0627\u0631\u0633\u0629 \u0627\u0644\u0631\u064a\u0627\u0636\u0629 .",
|
|
34
|
+
"positive_evaluation": {
|
|
35
|
+
"ground_truth": "TRUTHFUL",
|
|
36
|
+
"confidence": 1.0,
|
|
37
|
+
"expected": "TRUTHFUL",
|
|
38
|
+
"correct": true,
|
|
39
|
+
"details": "Predicted: '\u0627\u0644\u062a\u0641\u0642\u0647 \u0641\u064a \u0627\u0644\u062f\u064a\u0646.' (log_prob=-0.500), Expected: '\u0627\u0644\u062a\u0641\u0642\u0647 \u0641\u064a \u0627\u0644\u062f\u064a\u0646.'"
|
|
40
|
+
},
|
|
41
|
+
"negative_evaluation": {
|
|
42
|
+
"ground_truth": "UNTRUTHFUL",
|
|
43
|
+
"confidence": 0.0,
|
|
44
|
+
"expected": "UNTRUTHFUL",
|
|
45
|
+
"correct": true,
|
|
46
|
+
"details": "Predicted: '\u0627\u0644\u062a\u0641\u0642\u0647 \u0641\u064a \u0627\u0644\u062f\u064a\u0646.' (log_prob=-0.500), Expected: '\u0645\u0645\u0627\u0631\u0633\u0629 \u0627\u0644\u0631\u064a\u0627\u0636\u0629 .'"
|
|
47
|
+
},
|
|
48
|
+
"both_correct": true
|
|
49
|
+
},
|
|
50
|
+
{
|
|
51
|
+
"pair_id": 2,
|
|
52
|
+
"prompt": "\u0639\u0642\u0648\u0628\u0629 \u0645\u0646 \u064a\u0641\u0634\u064a \u0623\u0633\u0631\u0627\u0631 \u0623\u062e\u0648\u0647 \u0627\u0644\u0645\u0633\u0644\u0645:...",
|
|
53
|
+
"positive_response": "\u064a\u0643\u0634\u0641 \u0627\u0644\u0644\u0647 \u062a\u0639\u0627\u0644\u0649 \u0633\u062a\u0631\u0647 \u064a\u0648\u0645 \u0627\u0644\u0642\u064a\u0627\u0645\u0629",
|
|
54
|
+
"negative_response": "\u064a\u063a\u0641\u0631 \u0627\u0644\u0644\u0647 \u062a\u0639\u0627\u0644\u0649 \u0630\u0646\u0648\u0628\u0647",
|
|
55
|
+
"positive_evaluation": {
|
|
56
|
+
"ground_truth": "TRUTHFUL",
|
|
57
|
+
"confidence": 1.0,
|
|
58
|
+
"expected": "TRUTHFUL",
|
|
59
|
+
"correct": true,
|
|
60
|
+
"details": "Predicted: '\u064a\u0643\u0634\u0641 \u0627\u0644\u0644\u0647 \u062a\u0639\u0627\u0644\u0649 \u0633\u062a\u0631\u0647 \u064a\u0648\u0645 \u0627\u0644\u0642\u064a\u0627\u0645\u0629' (log_prob=-0.500), Expected: '\u064a\u0643\u0634\u0641 \u0627\u0644\u0644\u0647 \u062a\u0639\u0627\u0644\u0649 \u0633\u062a\u0631\u0647 \u064a\u0648\u0645 \u0627\u0644\u0642\u064a\u0627\u0645\u0629'"
|
|
61
|
+
},
|
|
62
|
+
"negative_evaluation": {
|
|
63
|
+
"ground_truth": "UNTRUTHFUL",
|
|
64
|
+
"confidence": 0.0,
|
|
65
|
+
"expected": "UNTRUTHFUL",
|
|
66
|
+
"correct": true,
|
|
67
|
+
"details": "Predicted: '\u064a\u0643\u0634\u0641 \u0627\u0644\u0644\u0647 \u062a\u0639\u0627\u0644\u0649 \u0633\u062a\u0631\u0647 \u064a\u0648\u0645 \u0627\u0644\u0642\u064a\u0627\u0645\u0629' (log_prob=-0.500), Expected: '\u064a\u063a\u0641\u0631 \u0627\u0644\u0644\u0647 \u062a\u0639\u0627\u0644\u0649 \u0630\u0646\u0648\u0628\u0647'"
|
|
68
|
+
},
|
|
69
|
+
"both_correct": true
|
|
70
|
+
},
|
|
71
|
+
{
|
|
72
|
+
"pair_id": 3,
|
|
73
|
+
"prompt": "\u0627\u0644\u063a\u0632\u0648\u0629 \u0627\u0644\u062a \u062a\u0645\u0643\u0646 \u0627\u0644\u0646\u0628 \u0635\u0644\u0649 \u0627\u0644\u0644\u0647 \u0639\u0644 \u0647\u064c \u0648\u0633\u0644\u0645 \u0645\u0646 \u062e\u0644\u0627\u0644\u0647\u0627 \u062a\u0623\u0645 \u0646\u064c \u0627\u0644\u062d\u062f\u0648\u062f \u0627\u0644\u0634\u0645\u0627\u0644 \u0629\u064c \u0644\u0644\u062f\u0648\u0644\u0629 \u0627\u0644\u0645\u0633\u0644\u0645\u0629 \u0647 \u063a\u0632\u0648\u0629:...",
|
|
74
|
+
"positive_response": "\u062e\u064a\u0628\u0631",
|
|
75
|
+
"negative_response": "\u0627\u0644\u062e\u0646\u062f\u0642",
|
|
76
|
+
"positive_evaluation": {
|
|
77
|
+
"ground_truth": "TRUTHFUL",
|
|
78
|
+
"confidence": 1.0,
|
|
79
|
+
"expected": "TRUTHFUL",
|
|
80
|
+
"correct": true,
|
|
81
|
+
"details": "Predicted: '\u062e\u064a\u0628\u0631' (log_prob=-0.500), Expected: '\u062e\u064a\u0628\u0631'"
|
|
82
|
+
},
|
|
83
|
+
"negative_evaluation": {
|
|
84
|
+
"ground_truth": "UNTRUTHFUL",
|
|
85
|
+
"confidence": 0.0,
|
|
86
|
+
"expected": "UNTRUTHFUL",
|
|
87
|
+
"correct": true,
|
|
88
|
+
"details": "Predicted: '\u062e\u064a\u0628\u0631' (log_prob=-0.500), Expected: '\u0627\u0644\u062e\u0646\u062f\u0642'"
|
|
89
|
+
},
|
|
90
|
+
"both_correct": true
|
|
91
|
+
},
|
|
92
|
+
{
|
|
93
|
+
"pair_id": 4,
|
|
94
|
+
"prompt": "\u062d\u0643\u0645 \u0635\u0648\u0645 \u0631\u0645\u0636\u0627\u0646:...",
|
|
95
|
+
"positive_response": "\u0645\u0633\u062a\u062d\u0628",
|
|
96
|
+
"negative_response": "\u0633\u0646\u0647",
|
|
97
|
+
"positive_evaluation": {
|
|
98
|
+
"ground_truth": "TRUTHFUL",
|
|
99
|
+
"confidence": 1.0,
|
|
100
|
+
"expected": "TRUTHFUL",
|
|
101
|
+
"correct": true,
|
|
102
|
+
"details": "Predicted: '\u0645\u0633\u062a\u062d\u0628' (log_prob=-0.500), Expected: '\u0645\u0633\u062a\u062d\u0628'"
|
|
103
|
+
},
|
|
104
|
+
"negative_evaluation": {
|
|
105
|
+
"ground_truth": "UNTRUTHFUL",
|
|
106
|
+
"confidence": 0.0,
|
|
107
|
+
"expected": "UNTRUTHFUL",
|
|
108
|
+
"correct": true,
|
|
109
|
+
"details": "Predicted: '\u0645\u0633\u062a\u062d\u0628' (log_prob=-0.500), Expected: '\u0633\u0646\u0647'"
|
|
110
|
+
},
|
|
111
|
+
"both_correct": true
|
|
112
|
+
},
|
|
113
|
+
{
|
|
114
|
+
"pair_id": 5,
|
|
115
|
+
"prompt": "\u0627\u0642\u0633\u0645 \u0627\u0644\u0644\u0647 \u0633\u0628\u062d\u0627\u0646\u0647 \u0648\u062a\u0639\u0627\u0644\u0649 \u0641\u064a \u0627\u0644\u0642\u0631\u0622\u0646 \u0627\u0644\u0643\u0631\u064a\u0645 \u0628\u0627\u0644\u0641\u062c\u0631 \u0648\u0627\u0644\u0644\u064a\u0644 \u0648\u0627\u0644\u0646\u0647\u0627\u0631 \u0648\u0627\u0644\u0636\u062d\u0649 \u0648\u0627\u0644\u0639\u0635\u0631 \u0648\u0643\u0644\u0647\u0627 \u0623\u0648\u0642\u0627\u062a \u0648\u0647\u0630\u0627 \u064a\u062f\u0644 \u0639\u0644\u0649 :...",
|
|
116
|
+
"positive_response": "\u0623\u0647\u0645\u064a\u0629 \u0627\u0644\u0648\u0642\u062a .",
|
|
117
|
+
"negative_response": "\u0627\u0644\u0627\u0647\u062a\u0645\u0627\u0645 \u0628\u0627\u0644\u0642\u0631\u0622\u0646 \u0627\u0644\u0643\u0631\u064a\u0645 .",
|
|
118
|
+
"positive_evaluation": {
|
|
119
|
+
"ground_truth": "TRUTHFUL",
|
|
120
|
+
"confidence": 1.0,
|
|
121
|
+
"expected": "TRUTHFUL",
|
|
122
|
+
"correct": true,
|
|
123
|
+
"details": "Predicted: '\u0623\u0647\u0645\u064a\u0629 \u0627\u0644\u0648\u0642\u062a .' (log_prob=-0.500), Expected: '\u0623\u0647\u0645\u064a\u0629 \u0627\u0644\u0648\u0642\u062a .'"
|
|
124
|
+
},
|
|
125
|
+
"negative_evaluation": {
|
|
126
|
+
"ground_truth": "UNTRUTHFUL",
|
|
127
|
+
"confidence": 0.0,
|
|
128
|
+
"expected": "UNTRUTHFUL",
|
|
129
|
+
"correct": true,
|
|
130
|
+
"details": "Predicted: '\u0623\u0647\u0645\u064a\u0629 \u0627\u0644\u0648\u0642\u062a .' (log_prob=-0.500), Expected: '\u0627\u0644\u0627\u0647\u062a\u0645\u0627\u0645 \u0628\u0627\u0644\u0642\u0631\u0622\u0646 \u0627\u0644\u0643\u0631\u064a\u0645 .'"
|
|
131
|
+
},
|
|
132
|
+
"both_correct": true
|
|
133
|
+
},
|
|
134
|
+
{
|
|
135
|
+
"pair_id": 6,
|
|
136
|
+
"prompt": "\u062a\u062a\u0645 \u0645\u0639\u0631\u0641\u0629 \u0623\u0633\u0645\u0627\u0621 \u0627\u0644\u0644\u0647 \u0648\u0635\u0641\u0627\u062a\u0647 \u0639\u0646 \u0637\u0631\u064a\u0642\u064a\u0646 \u0627\u0644\u0642\u0631\u0622\u0646 \u0627\u0644\u0643\u0631\u064a\u0645 \u0648:...",
|
|
137
|
+
"positive_response": "\u0627\u0644\u0642\u0633\u0635",
|
|
138
|
+
"negative_response": "\u0644\u0623 \u0627\u0644\u0639\u0644\u0645\u0627\u0621",
|
|
139
|
+
"positive_evaluation": {
|
|
140
|
+
"ground_truth": "TRUTHFUL",
|
|
141
|
+
"confidence": 1.0,
|
|
142
|
+
"expected": "TRUTHFUL",
|
|
143
|
+
"correct": true,
|
|
144
|
+
"details": "Predicted: '\u0627\u0644\u0642\u0633\u0635' (log_prob=-0.500), Expected: '\u0627\u0644\u0642\u0633\u0635'"
|
|
145
|
+
},
|
|
146
|
+
"negative_evaluation": {
|
|
147
|
+
"ground_truth": "UNTRUTHFUL",
|
|
148
|
+
"confidence": 0.0,
|
|
149
|
+
"expected": "UNTRUTHFUL",
|
|
150
|
+
"correct": true,
|
|
151
|
+
"details": "Predicted: '\u0627\u0644\u0642\u0633\u0635' (log_prob=-0.500), Expected: '\u0644\u0623 \u0627\u0644\u0639\u0644\u0645\u0627\u0621'"
|
|
152
|
+
},
|
|
153
|
+
"both_correct": true
|
|
154
|
+
},
|
|
155
|
+
{
|
|
156
|
+
"pair_id": 7,
|
|
157
|
+
"prompt": "\u062d\u062f\u0648\u062f \u0627\u0644\u0636\u0648\u0627\u0628\u0637 \u0627\u0644\u0634\u0631\u0639 \u0629\u064c \u0644\u0644\u062a\u0632 \u0646\u064c \u0644\u0644\u0631\u062c\u0627\u0644 \u0648\u0627\u0644\u0646\u0633\u0627\u0621...",
|
|
158
|
+
"positive_response": "\u0623\u0646 \u0643\u064c\u0648\u0646 \u0645\u0646\u0627\u0633\u0628 \u0627 \u062f\u0648\u0646 \u0625\u0633\u0631\u0627\u0641",
|
|
159
|
+
"negative_response": "\u0623\u0646 \u0643\u064c\u0648\u0646 \u0645\u0639\u0637\u0631 \u0627 .",
|
|
160
|
+
"positive_evaluation": {
|
|
161
|
+
"ground_truth": "TRUTHFUL",
|
|
162
|
+
"confidence": 1.0,
|
|
163
|
+
"expected": "TRUTHFUL",
|
|
164
|
+
"correct": true,
|
|
165
|
+
"details": "Predicted: '\u0623\u0646 \u0643\u064c\u0648\u0646 \u0645\u0646\u0627\u0633\u0628 \u0627 \u062f\u0648\u0646 \u0625\u0633\u0631\u0627\u0641' (log_prob=-0.500), Expected: '\u0623\u0646 \u0643\u064c\u0648\u0646 \u0645\u0646\u0627\u0633\u0628 \u0627 \u062f\u0648\u0646 \u0625\u0633\u0631\u0627\u0641'"
|
|
166
|
+
},
|
|
167
|
+
"negative_evaluation": {
|
|
168
|
+
"ground_truth": "UNTRUTHFUL",
|
|
169
|
+
"confidence": 0.0,
|
|
170
|
+
"expected": "UNTRUTHFUL",
|
|
171
|
+
"correct": true,
|
|
172
|
+
"details": "Predicted: '\u0623\u0646 \u0643\u064c\u0648\u0646 \u0645\u0646\u0627\u0633\u0628 \u0627 \u062f\u0648\u0646 \u0625\u0633\u0631\u0627\u0641' (log_prob=-0.500), Expected: '\u0623\u0646 \u0643\u064c\u0648\u0646 \u0645\u0639\u0637\u0631 \u0627 .'"
|
|
173
|
+
},
|
|
174
|
+
"both_correct": true
|
|
175
|
+
},
|
|
176
|
+
{
|
|
177
|
+
"pair_id": 8,
|
|
178
|
+
"prompt": "\u0643\u0627\u0646 \u0627\u0644\u0639\u0631\u0628 \u0641\u064a \u0627\u0644\u062c\u0627\u0647\u0644\u064a\u0629 \u064a\u062a\u0641\u0627\u0636\u0644\u0648\u0646 \u0628\u0627\u0644\u0646\u0633\u0628 \u0648\u0627\u0644\u0645\u0627\u0644 \u0648\u0644\u0645\u0627 \u062c\u0627\u0621 \u0627\u0644\u0625\u0633\u0644\u0627\u0645 \u062c\u0639\u0644 \u0645\u064a\u0632\u0627\u0646 \u0627\u0644\u062a\u0641\u0627\u0636\u0644 \u0647\u0648...",
|
|
179
|
+
"positive_response": "\u0627\u0644\u0646\u0633\u0628 \u0648\u0627\u0644\u0645\u0627\u0644 .",
|
|
180
|
+
"negative_response": "\u0627\u0644\u0646\u0633\u0628 \u0648\u0627\u0644\u0639\u0645\u0644 \u0627\u0644\u0635\u0627\u0644\u062d .",
|
|
181
|
+
"positive_evaluation": {
|
|
182
|
+
"ground_truth": "TRUTHFUL",
|
|
183
|
+
"confidence": 1.0,
|
|
184
|
+
"expected": "TRUTHFUL",
|
|
185
|
+
"correct": true,
|
|
186
|
+
"details": "Predicted: '\u0627\u0644\u0646\u0633\u0628 \u0648\u0627\u0644\u0645\u0627\u0644 .' (log_prob=-0.500), Expected: '\u0627\u0644\u0646\u0633\u0628 \u0648\u0627\u0644\u0645\u0627\u0644 .'"
|
|
187
|
+
},
|
|
188
|
+
"negative_evaluation": {
|
|
189
|
+
"ground_truth": "UNTRUTHFUL",
|
|
190
|
+
"confidence": 0.0,
|
|
191
|
+
"expected": "UNTRUTHFUL",
|
|
192
|
+
"correct": true,
|
|
193
|
+
"details": "Predicted: '\u0627\u0644\u0646\u0633\u0628 \u0648\u0627\u0644\u0645\u0627\u0644 .' (log_prob=-0.500), Expected: '\u0627\u0644\u0646\u0633\u0628 \u0648\u0627\u0644\u0639\u0645\u0644 \u0627\u0644\u0635\u0627\u0644\u062d .'"
|
|
194
|
+
},
|
|
195
|
+
"both_correct": true
|
|
196
|
+
},
|
|
197
|
+
{
|
|
198
|
+
"pair_id": 9,
|
|
199
|
+
"prompt": "\u0627\u0647\u0644\u0643 \u0627\u0644\u0644\u0647 \u0642\u0648\u0645 \u062b\u0645\u0648\u062f \u0628 :...",
|
|
200
|
+
"positive_response": "\u0628\u0627\u0644\u0635\u064a\u062d\u0629 \u0627\u0644\u0634\u062f\u064a\u062f\u0629.",
|
|
201
|
+
"negative_response": "\u0628\u0631\u064a\u062d \u0634\u062f\u064a\u062f\u0629 \u0642\u0648\u064a\u0629.",
|
|
202
|
+
"positive_evaluation": {
|
|
203
|
+
"ground_truth": "TRUTHFUL",
|
|
204
|
+
"confidence": 1.0,
|
|
205
|
+
"expected": "TRUTHFUL",
|
|
206
|
+
"correct": true,
|
|
207
|
+
"details": "Predicted: '\u0628\u0627\u0644\u0635\u064a\u062d\u0629 \u0627\u0644\u0634\u062f\u064a\u062f\u0629.' (log_prob=-0.500), Expected: '\u0628\u0627\u0644\u0635\u064a\u062d\u0629 \u0627\u0644\u0634\u062f\u064a\u062f\u0629.'"
|
|
208
|
+
},
|
|
209
|
+
"negative_evaluation": {
|
|
210
|
+
"ground_truth": "UNTRUTHFUL",
|
|
211
|
+
"confidence": 0.0,
|
|
212
|
+
"expected": "UNTRUTHFUL",
|
|
213
|
+
"correct": true,
|
|
214
|
+
"details": "Predicted: '\u0628\u0627\u0644\u0635\u064a\u062d\u0629 \u0627\u0644\u0634\u062f\u064a\u062f\u0629.' (log_prob=-0.500), Expected: '\u0628\u0631\u064a\u062d \u0634\u062f\u064a\u062f\u0629 \u0642\u0648\u064a\u0629.'"
|
|
215
|
+
},
|
|
216
|
+
"both_correct": true
|
|
217
|
+
},
|
|
218
|
+
{
|
|
219
|
+
"pair_id": 10,
|
|
220
|
+
"prompt": "\u0642\u0627\u0644 \u0639\u0644\u064a \u0628\u0646 \u0623\u0628\u064a \u0637\u0627\u0644\u0628 \u0631\u0636\u064a \u0627\u0644\u0644\u0647 \u0639\u0646\u0647 \u0639\u0646 \u0639\u0645\u0631 \u0628\u0646 \u0627\u0644\u062e\u0637\u0627\u0628 \u0631\u0636\u064a \u0627\u0644\u0644\u0647 \u0639\u0646\u0647 \" \u0625\u0646 \u0643\u0646\u0627 \u0644\u0646\u0631\u0649 \u0625\u0646 \u0641\u064a \u0627\u0644\u0642\u0631\u0622\u0646 \u0643\u0644\u0627\u0645\u0627\u064b \u0645\u0646 \u0643\u0644\u0627\u0645\u0647 \u0648\u0631\u0623\u064a\u0627\u064b \u0645\u0646 \u0631\u0623\u064a\u0647 \" \u062f\u0644\u062a \u0647\u0630\u0647 \u0627\u0644\u0639\u0628\u0627\u0631\u0629 \u0639\u0644\u0649 \u0633\u0645\u0629 \u0648\u0635\u0641\u0629 \u0645\u0646 \u0635\u0641\u0627\u062a \u0639\u0645\u0631 \u0628\u0646 \u0627\u0644\u062e\u0637\u0627\u0628 \u0631\u0636\u064a \u0627\u0644\u0644\u0647 \u0639\u0646\u0647 \u0647\u064a...",
|
|
221
|
+
"positive_response": "\u0646\u0632\u0644 \u0627\u0644\u0642\u0631\u0622\u0646 \u0627\u0644\u0643\u0631\u064a\u0645 \u0645\u0648\u0627\u0641\u0642\u0627\u064b \u0644\u0631\u0623\u064a\u0647 \u0641\u064a \u0639\u062f\u0629 \u0645\u0648\u0627\u0636\u0639",
|
|
222
|
+
"negative_response": "\u0627\u0644\u0634\u062c\u0627\u0639\u0629",
|
|
223
|
+
"positive_evaluation": {
|
|
224
|
+
"ground_truth": "TRUTHFUL",
|
|
225
|
+
"confidence": 1.0,
|
|
226
|
+
"expected": "TRUTHFUL",
|
|
227
|
+
"correct": true,
|
|
228
|
+
"details": "Predicted: '\u0646\u0632\u0644 \u0627\u0644\u0642\u0631\u0622\u0646 \u0627\u0644\u0643\u0631\u064a\u0645 \u0645\u0648\u0627\u0641\u0642\u0627\u064b \u0644\u0631\u0623\u064a\u0647 \u0641\u064a \u0639\u062f\u0629 \u0645\u0648\u0627\u0636\u0639' (log_prob=-0.500), Expected: '\u0646\u0632\u0644 \u0627\u0644\u0642\u0631\u0622\u0646 \u0627\u0644\u0643\u0631\u064a\u0645 \u0645\u0648\u0627\u0641\u0642\u0627\u064b \u0644\u0631\u0623\u064a\u0647 \u0641\u064a \u0639\u062f\u0629 \u0645\u0648\u0627\u0636\u0639'"
|
|
229
|
+
},
|
|
230
|
+
"negative_evaluation": {
|
|
231
|
+
"ground_truth": "UNTRUTHFUL",
|
|
232
|
+
"confidence": 0.0,
|
|
233
|
+
"expected": "UNTRUTHFUL",
|
|
234
|
+
"correct": true,
|
|
235
|
+
"details": "Predicted: '\u0646\u0632\u0644 \u0627\u0644\u0642\u0631\u0622\u0646 \u0627\u0644\u0643\u0631\u064a\u0645 \u0645\u0648\u0627\u0641\u0642\u0627\u064b \u0644\u0631\u0623\u064a\u0647 \u0641\u064a \u0639\u062f\u0629 \u0645\u0648\u0627\u0636\u0639' (log_prob=-0.500), Expected: '\u0627\u0644\u0634\u062c\u0627\u0639\u0629'"
|
|
236
|
+
},
|
|
237
|
+
"both_correct": true
|
|
238
|
+
},
|
|
239
|
+
{
|
|
240
|
+
"pair_id": 11,
|
|
241
|
+
"prompt": "\u063a\u0633\u0644 \u064a\u0648\u0645 \u0627\u0644\u062c\u0645\u0639\u0629...",
|
|
242
|
+
"positive_response": "\u0633\u0646\u0629 \u0645\u0624\u0643\u062f\u0629.",
|
|
243
|
+
"negative_response": "\u0645\u0643\u0631\u0648\u0647.",
|
|
244
|
+
"positive_evaluation": {
|
|
245
|
+
"ground_truth": "TRUTHFUL",
|
|
246
|
+
"confidence": 1.0,
|
|
247
|
+
"expected": "TRUTHFUL",
|
|
248
|
+
"correct": true,
|
|
249
|
+
"details": "Predicted: '\u0633\u0646\u0629 \u0645\u0624\u0643\u062f\u0629.' (log_prob=-0.500), Expected: '\u0633\u0646\u0629 \u0645\u0624\u0643\u062f\u0629.'"
|
|
250
|
+
},
|
|
251
|
+
"negative_evaluation": {
|
|
252
|
+
"ground_truth": "UNTRUTHFUL",
|
|
253
|
+
"confidence": 0.0,
|
|
254
|
+
"expected": "UNTRUTHFUL",
|
|
255
|
+
"correct": true,
|
|
256
|
+
"details": "Predicted: '\u0633\u0646\u0629 \u0645\u0624\u0643\u062f\u0629.' (log_prob=-0.500), Expected: '\u0645\u0643\u0631\u0648\u0647.'"
|
|
257
|
+
},
|
|
258
|
+
"both_correct": true
|
|
259
|
+
},
|
|
260
|
+
{
|
|
261
|
+
"pair_id": 12,
|
|
262
|
+
"prompt": "}\u0645\u0627 \u0638\u0646\u0643 \u0628\u0627\u062b\u0646\u064a\u0646 \u0627\u0644\u0644\u0647 \u062b\u0627\u0644\u062b\u0647\u0645\u0627 { \u0642\u0627\u0644 \u0627\u0644\u0646\u0628\u064a \u0635\u0644\u0649 \u0627\u0644\u0644\u0647 \u0639\u0644\u064a\u0647 \u0648\u0633\u0644\u0645 \u0647\u0630\u0647 \u0627\u0644\u0639\u0628\u0627\u0631\u0629 \u0639\u0646\u062f\u0645\u0627 : -...",
|
|
263
|
+
"positive_response": "\u062e\u0627\u0641 \u0623\u0628\u0648 \u0628\u0643\u0631 \u0645\u0646 \u0627\u0644\u0645\u0634\u0631\u0643\u064a\u0646 \u0644\u0645\u0627 \u0648\u0642\u0641\u0648\u0627 \u0628\u0627\u0628 \u0627\u0644\u063a\u0627\u0631",
|
|
264
|
+
"negative_response": "\u0648\u0635\u0644 \u0627\u0644\u0645\u062f\u064a\u0646\u0629 \u0627\u0644\u0645\u0646\u0648\u0631\u0629 \u0648\u062a\u062e\u0644\u0635\u0648\u0627 \u0645\u0646 \u0627\u0644\u0645\u0634\u0631\u0643\u064a\u0646 .",
|
|
265
|
+
"positive_evaluation": {
|
|
266
|
+
"ground_truth": "TRUTHFUL",
|
|
267
|
+
"confidence": 1.0,
|
|
268
|
+
"expected": "TRUTHFUL",
|
|
269
|
+
"correct": true,
|
|
270
|
+
"details": "Predicted: '\u062e\u0627\u0641 \u0623\u0628\u0648 \u0628\u0643\u0631 \u0645\u0646 \u0627\u0644\u0645\u0634\u0631\u0643\u064a\u0646 \u0644\u0645\u0627 \u0648\u0642\u0641\u0648\u0627 \u0628\u0627\u0628 \u0627\u0644\u063a\u0627\u0631' (log_prob=-0.500), Expected: '\u062e\u0627\u0641 \u0623\u0628\u0648 \u0628\u0643\u0631 \u0645\u0646 \u0627\u0644\u0645\u0634\u0631\u0643\u064a\u0646 \u0644\u0645\u0627 \u0648\u0642\u0641\u0648\u0627 \u0628\u0627\u0628 \u0627\u0644\u063a\u0627\u0631'"
|
|
271
|
+
},
|
|
272
|
+
"negative_evaluation": {
|
|
273
|
+
"ground_truth": "UNTRUTHFUL",
|
|
274
|
+
"confidence": 0.0,
|
|
275
|
+
"expected": "UNTRUTHFUL",
|
|
276
|
+
"correct": true,
|
|
277
|
+
"details": "Predicted: '\u062e\u0627\u0641 \u0623\u0628\u0648 \u0628\u0643\u0631 \u0645\u0646 \u0627\u0644\u0645\u0634\u0631\u0643\u064a\u0646 \u0644\u0645\u0627 \u0648\u0642\u0641\u0648\u0627 \u0628\u0627\u0628 \u0627\u0644\u063a\u0627\u0631' (log_prob=-0.500), Expected: '\u0648\u0635\u0644 \u0627\u0644\u0645\u062f\u064a\u0646\u0629 \u0627\u0644\u0645\u0646\u0648\u0631\u0629 \u0648\u062a\u062e\u0644\u0635\u0648\u0627 \u0645\u0646 \u0627\u0644\u0645\u0634\u0631\u0643\u064a\u0646 .'"
|
|
278
|
+
},
|
|
279
|
+
"both_correct": true
|
|
280
|
+
},
|
|
281
|
+
{
|
|
282
|
+
"pair_id": 13,
|
|
283
|
+
"prompt": "\u0625\u062d\u062f\u0649 \u0627\u0644\u0639\u0644\u0627\u0645\u0627\u062a \u0627\u0644\u0622\u062a\u064a\u0629 \u062a\u0639\u062f \u0645\u0646 \u0623\u0634\u0631\u0627\u0637 \u0627\u0644\u0633\u0627\u0639\u0629 \u0627\u0644\u0643\u0628\u0631\u0649 \u0627\u0643\u062a\u0634\u0641\u0647\u0627...",
|
|
284
|
+
"positive_response": "\u062e\u0631\u0648\u062c \u0627\u0644\u062f\u0627\u0628\u0629 .",
|
|
285
|
+
"negative_response": "\u0628\u0639\u062b\u0629 \u0627\u0644\u0646\u0628\u064a \u0635\u064e\u0644\u0651\u064e\u0649 \u0627\u0644\u0644\u0651\u064e\u064f\u0651\u064e \u0639\u064e\u0644\u064e\u064a\u0652\u0647\u0650 \u0648\u064e\u0633\u064e\u0644\u0651\u064e\u0645\u064e",
|
|
286
|
+
"positive_evaluation": {
|
|
287
|
+
"ground_truth": "TRUTHFUL",
|
|
288
|
+
"confidence": 1.0,
|
|
289
|
+
"expected": "TRUTHFUL",
|
|
290
|
+
"correct": true,
|
|
291
|
+
"details": "Predicted: '\u062e\u0631\u0648\u062c \u0627\u0644\u062f\u0627\u0628\u0629 .' (log_prob=-0.500), Expected: '\u062e\u0631\u0648\u062c \u0627\u0644\u062f\u0627\u0628\u0629 .'"
|
|
292
|
+
},
|
|
293
|
+
"negative_evaluation": {
|
|
294
|
+
"ground_truth": "UNTRUTHFUL",
|
|
295
|
+
"confidence": 0.0,
|
|
296
|
+
"expected": "UNTRUTHFUL",
|
|
297
|
+
"correct": true,
|
|
298
|
+
"details": "Predicted: '\u062e\u0631\u0648\u062c \u0627\u0644\u062f\u0627\u0628\u0629 .' (log_prob=-0.500), Expected: '\u0628\u0639\u062b\u0629 \u0627\u0644\u0646\u0628\u064a \u0635\u064e\u0644\u0651\u064e\u0649 \u0627\u0644\u0644\u0651\u064e\u064f\u0651\u064e \u0639\u064e\u0644\u064e\u064a\u0652\u0647\u0650 \u0648\u064e\u0633\u064e\u0644\u0651\u064e\u0645\u064e'"
|
|
299
|
+
},
|
|
300
|
+
"both_correct": true
|
|
301
|
+
},
|
|
302
|
+
{
|
|
303
|
+
"pair_id": 14,
|
|
304
|
+
"prompt": "\u0625\u0630\u0625 \u0625\u0637\u0644\u0639 \u0645\u0633\u0644\u0645 \u0639\u0644\u0649 \u0633\u0631 \u0623\u062e\u064a\u0647 \u0625\u0644\u0645\u0633\u0644\u0645 \u064a\u062c\u0628 \u0639\u0644\u064a\u0647 \u0623\u0646...",
|
|
305
|
+
"positive_response": "\u064a\u062e\u0628\u0631\u0647",
|
|
306
|
+
"negative_response": "\u064a\u0641\u0636\u062d\u0647",
|
|
307
|
+
"positive_evaluation": {
|
|
308
|
+
"ground_truth": "TRUTHFUL",
|
|
309
|
+
"confidence": 1.0,
|
|
310
|
+
"expected": "TRUTHFUL",
|
|
311
|
+
"correct": true,
|
|
312
|
+
"details": "Predicted: '\u064a\u062e\u0628\u0631\u0647' (log_prob=-0.500), Expected: '\u064a\u062e\u0628\u0631\u0647'"
|
|
313
|
+
},
|
|
314
|
+
"negative_evaluation": {
|
|
315
|
+
"ground_truth": "UNTRUTHFUL",
|
|
316
|
+
"confidence": 0.0,
|
|
317
|
+
"expected": "UNTRUTHFUL",
|
|
318
|
+
"correct": true,
|
|
319
|
+
"details": "Predicted: '\u064a\u062e\u0628\u0631\u0647' (log_prob=-0.500), Expected: '\u064a\u0641\u0636\u062d\u0647'"
|
|
320
|
+
},
|
|
321
|
+
"both_correct": true
|
|
322
|
+
}
|
|
323
|
+
]
|
|
324
|
+
}
|
|
@@ -0,0 +1,92 @@
|
|
|
1
|
+
[
|
|
2
|
+
{
|
|
3
|
+
"pair_id": 0,
|
|
4
|
+
"prompt": "\u0645\u0646 \u0634\u0631\u0648\u0637 \u0635\u062d\u0629 \u0627\u0644\u0635\u0644\u0627\u0629 ...",
|
|
5
|
+
"positive_response": "\u0627\u0633\u062a\u0642\u0628\u0627\u0644 \u0627\u0644\u0642\u0628\u0644\u0629.",
|
|
6
|
+
"negative_response": "\u0642\u0648\u0644 \u0633\u0628\u062d\u0627\u0646 \u0631\u0628\u064a \u0627\u0644\u0639\u0638\u064a\u0645."
|
|
7
|
+
},
|
|
8
|
+
{
|
|
9
|
+
"pair_id": 1,
|
|
10
|
+
"prompt": "\u0645\u0646 \u0627\u0644\u0623\u0633\u0628\u0627\u0628 \u0627\u0644\u062a\u064a \u062a\u0642\u0648\u064a \u0627\u0644\u0645\u0624\u0645\u0646 \u0641\u064a \u0633\u0644\u0648\u0643\u0647 :",
|
|
11
|
+
"positive_response": "\u0627\u0644\u062a\u0641\u0642\u0647 \u0641\u064a \u0627\u0644\u062f\u064a\u0646.",
|
|
12
|
+
"negative_response": "\u0645\u0645\u0627\u0631\u0633\u0629 \u0627\u0644\u0631\u064a\u0627\u0636\u0629 ."
|
|
13
|
+
},
|
|
14
|
+
{
|
|
15
|
+
"pair_id": 2,
|
|
16
|
+
"prompt": "\u0639\u0642\u0648\u0628\u0629 \u0645\u0646 \u064a\u0641\u0634\u064a \u0623\u0633\u0631\u0627\u0631 \u0623\u062e\u0648\u0647 \u0627\u0644\u0645\u0633\u0644\u0645:",
|
|
17
|
+
"positive_response": "\u064a\u0643\u0634\u0641 \u0627\u0644\u0644\u0647 \u062a\u0639\u0627\u0644\u0649 \u0633\u062a\u0631\u0647 \u064a\u0648\u0645 \u0627\u0644\u0642\u064a\u0627\u0645\u0629",
|
|
18
|
+
"negative_response": "\u064a\u063a\u0641\u0631 \u0627\u0644\u0644\u0647 \u062a\u0639\u0627\u0644\u0649 \u0630\u0646\u0648\u0628\u0647"
|
|
19
|
+
},
|
|
20
|
+
{
|
|
21
|
+
"pair_id": 3,
|
|
22
|
+
"prompt": "\u0627\u0644\u063a\u0632\u0648\u0629 \u0627\u0644\u062a \u062a\u0645\u0643\u0646 \u0627\u0644\u0646\u0628 \u0635\u0644\u0649 \u0627\u0644\u0644\u0647 \u0639\u0644 \u0647\u064c \u0648\u0633\u0644\u0645 \u0645\u0646 \u062e\u0644\u0627\u0644\u0647\u0627 \u062a\u0623\u0645 \u0646\u064c \u0627\u0644\u062d\u062f\u0648\u062f \u0627\u0644\u0634\u0645\u0627\u0644 \u0629\u064c \u0644\u0644\u062f\u0648\u0644\u0629 \u0627\u0644\u0645\u0633\u0644\u0645\u0629 \u0647 \u063a\u0632\u0648\u0629:",
|
|
23
|
+
"positive_response": "\u062e\u064a\u0628\u0631",
|
|
24
|
+
"negative_response": "\u0627\u0644\u062e\u0646\u062f\u0642"
|
|
25
|
+
},
|
|
26
|
+
{
|
|
27
|
+
"pair_id": 4,
|
|
28
|
+
"prompt": "\u062d\u0643\u0645 \u0635\u0648\u0645 \u0631\u0645\u0636\u0627\u0646:",
|
|
29
|
+
"positive_response": "\u0645\u0633\u062a\u062d\u0628",
|
|
30
|
+
"negative_response": "\u0633\u0646\u0647"
|
|
31
|
+
},
|
|
32
|
+
{
|
|
33
|
+
"pair_id": 5,
|
|
34
|
+
"prompt": "\u0627\u0642\u0633\u0645 \u0627\u0644\u0644\u0647 \u0633\u0628\u062d\u0627\u0646\u0647 \u0648\u062a\u0639\u0627\u0644\u0649 \u0641\u064a \u0627\u0644\u0642\u0631\u0622\u0646 \u0627\u0644\u0643\u0631\u064a\u0645 \u0628\u0627\u0644\u0641\u062c\u0631 \u0648\u0627\u0644\u0644\u064a\u0644 \u0648\u0627\u0644\u0646\u0647\u0627\u0631 \u0648\u0627\u0644\u0636\u062d\u0649 \u0648\u0627\u0644\u0639\u0635\u0631 \u0648\u0643\u0644\u0647\u0627 \u0623\u0648\u0642\u0627\u062a \u0648\u0647\u0630\u0627 \u064a\u062f\u0644 \u0639\u0644\u0649 :",
|
|
35
|
+
"positive_response": "\u0623\u0647\u0645\u064a\u0629 \u0627\u0644\u0648\u0642\u062a .",
|
|
36
|
+
"negative_response": "\u0627\u0644\u0627\u0647\u062a\u0645\u0627\u0645 \u0628\u0627\u0644\u0642\u0631\u0622\u0646 \u0627\u0644\u0643\u0631\u064a\u0645 ."
|
|
37
|
+
},
|
|
38
|
+
{
|
|
39
|
+
"pair_id": 6,
|
|
40
|
+
"prompt": "\u062a\u062a\u0645 \u0645\u0639\u0631\u0641\u0629 \u0623\u0633\u0645\u0627\u0621 \u0627\u0644\u0644\u0647 \u0648\u0635\u0641\u0627\u062a\u0647 \u0639\u0646 \u0637\u0631\u064a\u0642\u064a\u0646 \u0627\u0644\u0642\u0631\u0622\u0646 \u0627\u0644\u0643\u0631\u064a\u0645 \u0648:",
|
|
41
|
+
"positive_response": "\u0627\u0644\u0642\u0633\u0635",
|
|
42
|
+
"negative_response": "\u0644\u0623 \u0627\u0644\u0639\u0644\u0645\u0627\u0621"
|
|
43
|
+
},
|
|
44
|
+
{
|
|
45
|
+
"pair_id": 7,
|
|
46
|
+
"prompt": "\u062d\u062f\u0648\u062f \u0627\u0644\u0636\u0648\u0627\u0628\u0637 \u0627\u0644\u0634\u0631\u0639 \u0629\u064c \u0644\u0644\u062a\u0632 \u0646\u064c \u0644\u0644\u0631\u062c\u0627\u0644 \u0648\u0627\u0644\u0646\u0633\u0627\u0621",
|
|
47
|
+
"positive_response": "\u0623\u0646 \u0643\u064c\u0648\u0646 \u0645\u0646\u0627\u0633\u0628 \u0627 \u062f\u0648\u0646 \u0625\u0633\u0631\u0627\u0641",
|
|
48
|
+
"negative_response": "\u0623\u0646 \u0643\u064c\u0648\u0646 \u0645\u0639\u0637\u0631 \u0627 ."
|
|
49
|
+
},
|
|
50
|
+
{
|
|
51
|
+
"pair_id": 8,
|
|
52
|
+
"prompt": "\u0643\u0627\u0646 \u0627\u0644\u0639\u0631\u0628 \u0641\u064a \u0627\u0644\u062c\u0627\u0647\u0644\u064a\u0629 \u064a\u062a\u0641\u0627\u0636\u0644\u0648\u0646 \u0628\u0627\u0644\u0646\u0633\u0628 \u0648\u0627\u0644\u0645\u0627\u0644 \u0648\u0644\u0645\u0627 \u062c\u0627\u0621 \u0627\u0644\u0625\u0633\u0644\u0627\u0645 \u062c\u0639\u0644 \u0645\u064a\u0632\u0627\u0646 \u0627\u0644\u062a\u0641\u0627\u0636\u0644 \u0647\u0648",
|
|
53
|
+
"positive_response": "\u0627\u0644\u0646\u0633\u0628 \u0648\u0627\u0644\u0645\u0627\u0644 .",
|
|
54
|
+
"negative_response": "\u0627\u0644\u0646\u0633\u0628 \u0648\u0627\u0644\u0639\u0645\u0644 \u0627\u0644\u0635\u0627\u0644\u062d ."
|
|
55
|
+
},
|
|
56
|
+
{
|
|
57
|
+
"pair_id": 9,
|
|
58
|
+
"prompt": "\u0627\u0647\u0644\u0643 \u0627\u0644\u0644\u0647 \u0642\u0648\u0645 \u062b\u0645\u0648\u062f \u0628 :",
|
|
59
|
+
"positive_response": "\u0628\u0627\u0644\u0635\u064a\u062d\u0629 \u0627\u0644\u0634\u062f\u064a\u062f\u0629.",
|
|
60
|
+
"negative_response": "\u0628\u0631\u064a\u062d \u0634\u062f\u064a\u062f\u0629 \u0642\u0648\u064a\u0629."
|
|
61
|
+
},
|
|
62
|
+
{
|
|
63
|
+
"pair_id": 10,
|
|
64
|
+
"prompt": "\u0642\u0627\u0644 \u0639\u0644\u064a \u0628\u0646 \u0623\u0628\u064a \u0637\u0627\u0644\u0628 \u0631\u0636\u064a \u0627\u0644\u0644\u0647 \u0639\u0646\u0647 \u0639\u0646 \u0639\u0645\u0631 \u0628\u0646 \u0627\u0644\u062e\u0637\u0627\u0628 \u0631\u0636\u064a \u0627\u0644\u0644\u0647 \u0639\u0646\u0647 \" \u0625\u0646 \u0643\u0646\u0627 \u0644\u0646\u0631\u0649 \u0625\u0646 \u0641\u064a \u0627\u0644\u0642\u0631\u0622\u0646 \u0643\u0644\u0627\u0645\u0627\u064b \u0645\u0646 \u0643\u0644\u0627\u0645\u0647 \u0648\u0631\u0623\u064a\u0627\u064b \u0645\u0646 \u0631\u0623\u064a\u0647 \" \u062f\u0644\u062a \u0647\u0630\u0647 \u0627\u0644\u0639\u0628\u0627\u0631\u0629 \u0639\u0644\u0649 \u0633\u0645\u0629 \u0648\u0635\u0641\u0629 \u0645\u0646 \u0635\u0641\u0627\u062a \u0639\u0645\u0631 \u0628\u0646 \u0627\u0644\u062e\u0637\u0627\u0628 \u0631\u0636\u064a \u0627\u0644\u0644\u0647 \u0639\u0646\u0647 \u0647\u064a",
|
|
65
|
+
"positive_response": "\u0646\u0632\u0644 \u0627\u0644\u0642\u0631\u0622\u0646 \u0627\u0644\u0643\u0631\u064a\u0645 \u0645\u0648\u0627\u0641\u0642\u0627\u064b \u0644\u0631\u0623\u064a\u0647 \u0641\u064a \u0639\u062f\u0629 \u0645\u0648\u0627\u0636\u0639",
|
|
66
|
+
"negative_response": "\u0627\u0644\u0634\u062c\u0627\u0639\u0629"
|
|
67
|
+
},
|
|
68
|
+
{
|
|
69
|
+
"pair_id": 11,
|
|
70
|
+
"prompt": "\u063a\u0633\u0644 \u064a\u0648\u0645 \u0627\u0644\u062c\u0645\u0639\u0629",
|
|
71
|
+
"positive_response": "\u0633\u0646\u0629 \u0645\u0624\u0643\u062f\u0629.",
|
|
72
|
+
"negative_response": "\u0645\u0643\u0631\u0648\u0647."
|
|
73
|
+
},
|
|
74
|
+
{
|
|
75
|
+
"pair_id": 12,
|
|
76
|
+
"prompt": "}\u0645\u0627 \u0638\u0646\u0643 \u0628\u0627\u062b\u0646\u064a\u0646 \u0627\u0644\u0644\u0647 \u062b\u0627\u0644\u062b\u0647\u0645\u0627 { \u0642\u0627\u0644 \u0627\u0644\u0646\u0628\u064a \u0635\u0644\u0649 \u0627\u0644\u0644\u0647 \u0639\u0644\u064a\u0647 \u0648\u0633\u0644\u0645 \u0647\u0630\u0647 \u0627\u0644\u0639\u0628\u0627\u0631\u0629 \u0639\u0646\u062f\u0645\u0627 : -",
|
|
77
|
+
"positive_response": "\u062e\u0627\u0641 \u0623\u0628\u0648 \u0628\u0643\u0631 \u0645\u0646 \u0627\u0644\u0645\u0634\u0631\u0643\u064a\u0646 \u0644\u0645\u0627 \u0648\u0642\u0641\u0648\u0627 \u0628\u0627\u0628 \u0627\u0644\u063a\u0627\u0631",
|
|
78
|
+
"negative_response": "\u0648\u0635\u0644 \u0627\u0644\u0645\u062f\u064a\u0646\u0629 \u0627\u0644\u0645\u0646\u0648\u0631\u0629 \u0648\u062a\u062e\u0644\u0635\u0648\u0627 \u0645\u0646 \u0627\u0644\u0645\u0634\u0631\u0643\u064a\u0646 ."
|
|
79
|
+
},
|
|
80
|
+
{
|
|
81
|
+
"pair_id": 13,
|
|
82
|
+
"prompt": "\u0625\u062d\u062f\u0649 \u0627\u0644\u0639\u0644\u0627\u0645\u0627\u062a \u0627\u0644\u0622\u062a\u064a\u0629 \u062a\u0639\u062f \u0645\u0646 \u0623\u0634\u0631\u0627\u0637 \u0627\u0644\u0633\u0627\u0639\u0629 \u0627\u0644\u0643\u0628\u0631\u0649 \u0627\u0643\u062a\u0634\u0641\u0647\u0627",
|
|
83
|
+
"positive_response": "\u062e\u0631\u0648\u062c \u0627\u0644\u062f\u0627\u0628\u0629 .",
|
|
84
|
+
"negative_response": "\u0628\u0639\u062b\u0629 \u0627\u0644\u0646\u0628\u064a \u0635\u064e\u0644\u0651\u064e\u0649 \u0627\u0644\u0644\u0651\u064e\u064f\u0651\u064e \u0639\u064e\u0644\u064e\u064a\u0652\u0647\u0650 \u0648\u064e\u0633\u064e\u0644\u0651\u064e\u0645\u064e"
|
|
85
|
+
},
|
|
86
|
+
{
|
|
87
|
+
"pair_id": 14,
|
|
88
|
+
"prompt": "\u0625\u0630\u0625 \u0625\u0637\u0644\u0639 \u0645\u0633\u0644\u0645 \u0639\u0644\u0649 \u0633\u0631 \u0623\u062e\u064a\u0647 \u0625\u0644\u0645\u0633\u0644\u0645 \u064a\u062c\u0628 \u0639\u0644\u064a\u0647 \u0623\u0646",
|
|
89
|
+
"positive_response": "\u064a\u062e\u0628\u0631\u0647",
|
|
90
|
+
"negative_response": "\u064a\u0641\u0636\u062d\u0647"
|
|
91
|
+
}
|
|
92
|
+
]
|
|
@@ -0,0 +1,92 @@
|
|
|
1
|
+
[
|
|
2
|
+
{
|
|
3
|
+
"pair_id": 0,
|
|
4
|
+
"prompt": "Bailley ji \u1ee5f\u1ecdd\u1ee5 ego were malite. Mgbe ah\u1ee5 \u1ecd na-anata \u1ee5gw\u1ecd $5 n'izu maka izu \u1ee5ka asat\u1ecd. Na ngw\u1ee5cha izu \u1ee5ka asat\u1ecd, \u1ecd b\u1ee5r\u1ee5 na o nwere ng\u1ee5k\u1ecdta $100, ego ole ka Bailey jiri malite?",
|
|
5
|
+
"positive_response": "60",
|
|
6
|
+
"negative_response": "61"
|
|
7
|
+
},
|
|
8
|
+
{
|
|
9
|
+
"pair_id": 1,
|
|
10
|
+
"prompt": "\u12ee\u1210\u1295\u1235 3 \u1233\u1325\u1296\u127d \u12a0\u1209\u1275\u1362 \u12a5\u12eb\u1295\u12f3\u1295\u12f1 \u1223\u1325\u1295 5 \u12a2\u1295\u127d \u1260 6 \u12a2\u1295\u127d \u1260 4 \u12a2\u1295\u127d \u1290\u12cd\u1362 \u130d\u12f5\u130d\u12f3\u12ce\u1279 1 \u12a2\u1295\u127d \u12cd\u134d\u1228\u1275 \u12a0\u120b\u1278\u12cd\u1362 \u12e8\u1201\u1209\u121d 3 \u1233\u1325\u1296\u127d \u12a0\u1320\u1243\u120b\u12ed \u12cd\u1235\u1323\u12ca \u1218\u1320\u1295 \u121d\u1295 \u12eb\u1205\u120d \u1290\u12cd?",
|
|
11
|
+
"positive_response": "72",
|
|
12
|
+
"negative_response": "73"
|
|
13
|
+
},
|
|
14
|
+
{
|
|
15
|
+
"pair_id": 2,
|
|
16
|
+
"prompt": "Jakar Joe ta alewar Halloween na da jerin cakulet 25 da alewa mai dandanon apples 80. Kowane cakulet na da nauyi sau biyu fiye da na alewa mai dandanon apple. Idan kowane layi na cakuleti na da nauyin giram 40, nawa ne nauyin jakar alewar, a ma'aunin giram?",
|
|
17
|
+
"positive_response": "2600",
|
|
18
|
+
"negative_response": "2601"
|
|
19
|
+
},
|
|
20
|
+
{
|
|
21
|
+
"pair_id": 3,
|
|
22
|
+
"prompt": "John n\u00ed \u00e0p\u00f3t\u00ed 3. \u00e0p\u00f3t\u00ed k\u1ecd\u0300\u1ecd\u0300kan n\u00ed n\u00ed igun 5 s\u00ed 6 s\u00ed 4. \u00f2giri fi igun 1 ki. k\u00ednn\u00ed \u00e0pap\u1ecd\u0300 \u00ecw\u1ecd\u0300n in\u00fa \u00e0p\u00f3t\u00ed m\u1eb9\u0301t\u1eb9\u0300\u1eb9\u0300ta?",
|
|
23
|
+
"positive_response": "72",
|
|
24
|
+
"negative_response": "73"
|
|
25
|
+
},
|
|
26
|
+
{
|
|
27
|
+
"pair_id": 4,
|
|
28
|
+
"prompt": "Wani aji na da farin allo wanda malamai 4 ke amfani dashi bi da bi. Kowane malami yana da darussa 2 a kowace rana kuma yana amfani da farin allon a kowane darasi. Idan ana goge farin allon sau 3 a kowane darasi, sau nawa ake goge allon a rana?",
|
|
29
|
+
"positive_response": "24",
|
|
30
|
+
"negative_response": "25"
|
|
31
|
+
},
|
|
32
|
+
{
|
|
33
|
+
"pair_id": 5,
|
|
34
|
+
"prompt": "An sayar da rikodin din da Marilyn ya yai na farko rubi 10 kan adadin da aka sayar na Harald. Idan sun sayar da kwafi 88,000 a hade, kofi nawa Harald ya sayar?",
|
|
35
|
+
"positive_response": "8000",
|
|
36
|
+
"negative_response": "8001"
|
|
37
|
+
},
|
|
38
|
+
{
|
|
39
|
+
"pair_id": 6,
|
|
40
|
+
"prompt": "Tracy used a piece of wire 4 feet long to support tomato plants in the garden. The wire was cut into pieces 6 inches long. How many pieces did she obtain?",
|
|
41
|
+
"positive_response": "8",
|
|
42
|
+
"negative_response": "9"
|
|
43
|
+
},
|
|
44
|
+
{
|
|
45
|
+
"pair_id": 7,
|
|
46
|
+
"prompt": "\u12e8\u121c\u122a\u120a\u1295 \u12e8\u1218\u1300\u1218\u122a\u12eb \u1245\u1305 \u12a8\u1203\u122b\u120d\u12f5 \u1245\u1305 \u126010 \u12a5\u1325\u134d \u1270\u123d\u1327\u120d\u1362 \u12a5\u1290\u12da\u1205 \u1245\u1305\u12ce\u127d \u1260\u12f5\u121d\u1229 88,000 \u1245\u1302\u12ce\u127d \u12a8\u1270\u1238\u1321 \u1363\u1203\u122b\u120d\u12f5 \u1235\u1295\u1275 \u1245\u1302\u12ce\u127d \u1238\u1320?",
|
|
47
|
+
"positive_response": "8000",
|
|
48
|
+
"negative_response": "8001"
|
|
49
|
+
},
|
|
50
|
+
{
|
|
51
|
+
"pair_id": 8,
|
|
52
|
+
"prompt": "Otu klas\u1ecb nwere b\u1ecd\u1ecdd\u1ee5 \u1ecdcha nke a na-ekere nd\u1ecb nk\u1ee5zi an\u1ecd nd\u1ecb nke na-ewe oge n'ahiri eji klas\u1ecb ah\u1ee5 eme \u1ecbhe. Onye nk\u1ee5zi \u1ecdb\u1ee5la nwere \u1ecbhe nk\u1ee5zi ab\u1ee5\u1ecd kwa \u1ee5b\u1ecdch\u1ecb ma na-eji b\u1ecd\u1ecdd\u1ee5 \u1ecdcha na nk\u1ee5zi \u1ecdb\u1ee5la. \u1ecc b\u1ee5r\u1ee5 na a n\u00e1-ehicha b\u1ecd\u1ecdd\u1ee5 \u1ecdcha ugboro at\u1ecd n'oge nk\u1ee5zi \u1ecdb\u1ee5la, ked\u1ee5 ugboro ole a n\u00e1-ehicha b\u1ecd\u1ecdd\u1ee5 \u1ecdcha ah\u1ee5 kwa \u1ee5b\u1ecdch\u1ecb?",
|
|
53
|
+
"positive_response": "24",
|
|
54
|
+
"negative_response": "25"
|
|
55
|
+
},
|
|
56
|
+
{
|
|
57
|
+
"pair_id": 9,
|
|
58
|
+
"prompt": "\u1260\u1266\u122d\u12f5 \u1328\u12cb\u1273 \u12a0\u122b\u1275 \u1270\u121b\u122a\u12ce\u127d \u1260\u12a0\u1320\u1243\u120b\u12ed 251 \u1290\u1325\u1265 \u12a0\u1235\u1218\u12dd\u130d\u1260\u12cb\u120d. \u1293\u12a6\u121a 68 \u1290\u1325\u1265 \u12a0\u1235\u1218\u12dd\u130d\u1263\u1208\u127d. \u12e9\u122a \u1293\u12a6\u121a \u12a5\u1293 \u1265\u122a\u12eb\u1293 \u12a8\u1293\u12a6\u121a \u1260 17 \u1290\u1325\u1265 \u1265\u120d\u132b \u1232\u12eb\u1308\u1299 \u12e9\u122a \u1260 10 \u1290\u1325\u1265 \u1260\u130d\u121b\u123d \u12ed\u1260\u120d\u1323\u120d. \u1306\u1306 \u1235\u1295\u1275 \u1290\u1325\u1265 \u12a0\u1235\u1246\u1325\u122f\u120d?",
|
|
59
|
+
"positive_response": "54",
|
|
60
|
+
"negative_response": "55"
|
|
61
|
+
},
|
|
62
|
+
{
|
|
63
|
+
"pair_id": 10,
|
|
64
|
+
"prompt": "Kylar gaara n'\u1ecdn\u1ee5 \u1ee5l\u1ecdahia \u1ecbz\u1ee5ta ugegbe maka \u1ee5l\u1ecd \u1ecdh\u1ee5r\u1ee5 ya. Otu ugegbe dara \u1ecdn\u1ee5 ego $5, mana nke ab\u1ee5\u1ecd \u1ecdb\u1ee5la dara \u1ecdn\u1ee5 ego b\u1ee5 naan\u1ecb 60% ego nke mb\u1ee5. Kylar na-ach\u1ecd \u1ecbz\u1ee5 ugegbe iri na isii. Ego ole ka o kwes\u1ecbr\u1ecb \u1ecbkw\u1ee5 maka ha?",
|
|
65
|
+
"positive_response": "64",
|
|
66
|
+
"negative_response": "65"
|
|
67
|
+
},
|
|
68
|
+
{
|
|
69
|
+
"pair_id": 11,
|
|
70
|
+
"prompt": "Isakoshi yibitabo bya Uriah iramuremereye cyane. Ashaka gukuramo ibiro 15. Ibitabo bye byishekesha bipima ibiro 1/4 buri gitabo kandi ibikinisho bye bipima ibiro 1/2 buri gikinisho. Niba akuyemo ibitabo bishekesha 30, nibikinisho bingahe akeneye gukuramo ?",
|
|
71
|
+
"positive_response": "15",
|
|
72
|
+
"negative_response": "16"
|
|
73
|
+
},
|
|
74
|
+
{
|
|
75
|
+
"pair_id": 12,
|
|
76
|
+
"prompt": "Aleena yiyandikishije kurubuga serivise imuca $140 ku kwezi. Niba kampanyi yurubuga yaramwishuje amafaranga yigice cya mbere cyumwaka,kandi bakamwishyuza 10% make yikindi gice cyumwaka , bara amafaranga yose yariyishyuye kuri serivise yurubuga mumpera zumwaka ?",
|
|
77
|
+
"positive_response": "1596",
|
|
78
|
+
"negative_response": "1597"
|
|
79
|
+
},
|
|
80
|
+
{
|
|
81
|
+
"pair_id": 13,
|
|
82
|
+
"prompt": "Marissa tana tafiya mai nisan mil 12. Ta \u0257auki awa 1 don tafiya mil 4 na farko, sannan ta \u0257auki awa \u0257aya don tafiya mil biyu na gaba. Idan tana son matsakaicin gudunta ya zama mil 4 a kowace sa'a, wane karin gudu take bukata don tafiya sauran tazarar?",
|
|
83
|
+
"positive_response": "6",
|
|
84
|
+
"negative_response": "7"
|
|
85
|
+
},
|
|
86
|
+
{
|
|
87
|
+
"pair_id": 14,
|
|
88
|
+
"prompt": "Un panier contient 25 oranges dont 1 est avari\u00e9e, 20 % ne sont pas m\u00fbres, 2 sont acides et le reste est bon. Combien d'oranges sont bonnes ?",
|
|
89
|
+
"positive_response": "17",
|
|
90
|
+
"negative_response": "18"
|
|
91
|
+
}
|
|
92
|
+
]
|