Update README with YAML configs for all 113 datasets

README.md

@@ -1,321 +1,466 @@
 ---
-license: apache-2.0
-task_categories:
-- text-classification
-language:
-- en
-size_categories:
-- 1K<n<10K
-tags:
-- probing
-- interpretability
-- sparse-autoencoders
-- nlp
 configs:
+- config_name: 2_hist_fig_birthyear
+  data_files:
+  - split: train
+    path: 2_hist_fig_birthyear/train-00000-of-00001.parquet
+- config_name: 3_hist_fig_deathyear
+  data_files:
+  - split: train
+    path: 3_hist_fig_deathyear/train-00000-of-00001.parquet
+- config_name: 4_hist_fig_age
+  data_files:
+  - split: train
+    path: 4_hist_fig_age/train-00000-of-00001.parquet
+- config_name: 5_hist_fig_ismale
+  data_files:
+  - split: train
+    path: 5_hist_fig_ismale/train-00000-of-00001.parquet
+- config_name: 6_hist_fig_isamerican
+  data_files:
+  - split: train
+    path: 6_hist_fig_isamerican/train-00000-of-00001.parquet
+- config_name: 7_hist_fig_ispolitician
+  data_files:
+  - split: train
+    path: 7_hist_fig_ispolitician/train-00000-of-00001.parquet
+- config_name: 8_nyc_lat
+  data_files:
+  - split: train
+    path: 8_nyc_lat/train-00000-of-00001.parquet
+- config_name: 9_nyc_long
+  data_files:
+  - split: train
+    path: 9_nyc_long/train-00000-of-00001.parquet
 - config_name: 10_nyc_borough
   data_files:
   - split: train
-    path: 10_nyc_borough/train-*
+    path: 10_nyc_borough/train-00000-of-00001.parquet
 - config_name: 11_us_lat
   data_files:
   - split: train
-    path: 11_us_lat/train-*
+    path: 11_us_lat/train-00000-of-00001.parquet
 - config_name: 12_us_long
   data_files:
   - split: train
-    path: 12_us_long/train-*
+    path: 12_us_long/train-00000-of-00001.parquet
 - config_name: 13_us_state
   data_files:
   - split: train
-    path: 13_us_state/train-*
+    path: 13_us_state/train-00000-of-00001.parquet
 - config_name: 14_us_timezone
   data_files:
   - split: train
-    path: 14_us_timezone/train-*
+    path: 14_us_timezone/train-00000-of-00001.parquet
 - config_name: 15_us_population
   data_files:
   - split: train
-    path: 15_us_population/train-*
+    path: 15_us_population/train-00000-of-00001.parquet
 - config_name: 16_us_density
   data_files:
   - split: train
-    path: 16_us_density/train-*
+    path: 16_us_density/train-00000-of-00001.parquet
 - config_name: 17_world_country
   data_files:
   - split: train
-    path: 17_world_country/train-*
+    path: 17_world_country/train-00000-of-00001.parquet
 - config_name: 18_world_latitude
   data_files:
   - split: train
-    path: 18_world_latitude/train-*
+    path: 18_world_latitude/train-00000-of-00001.parquet
 - config_name: 19_world_longitude
   data_files:
   - split: train
-    path: 19_world_longitude/train-*
-- config_name: 2_hist_fig_birthyear
+    path: 19_world_longitude/train-00000-of-00001.parquet
+- config_name: 20_world_pageviews
   data_files:
   - split: train
-    path: 2_hist_fig_birthyear/train-*
-- config_name: 3_hist_fig_deathyear
+    path: 20_world_pageviews/train-00000-of-00001.parquet
+- config_name: 21_headline_istrump
   data_files:
   - split: train
-    path: 3_hist_fig_deathyear/train-*
-- config_name: 4_hist_fig_age
+    path: 21_headline_istrump/train-00000-of-00001.parquet
+- config_name: 22_headline_isobama
   data_files:
   - split: train
-    path: 4_hist_fig_age/train-*
-- config_name: 5_hist_fig_ismale
+    path: 22_headline_isobama/train-00000-of-00001.parquet
+- config_name: 23_headline_ischina
   data_files:
   - split: train
-    path: 5_hist_fig_ismale/train-*
-- config_name: 6_hist_fig_isamerican
+    path: 23_headline_ischina/train-00000-of-00001.parquet
+- config_name: 24_headline_isiran
   data_files:
   - split: train
-    path: 6_hist_fig_isamerican/train-*
-- config_name: 7_hist_fig_ispolitician
+    path: 24_headline_isiran/train-00000-of-00001.parquet
+- config_name: 25_headline_year
   data_files:
   - split: train
-    path: 7_hist_fig_ispolitician/train-*
-- config_name: 8_nyc_lat
+    path: 25_headline_year/train-00000-of-00001.parquet
+- config_name: 26_headline_isfrontpage
   data_files:
   - split: train
-    path: 8_nyc_lat/train-*
-- config_name: 9_nyc_long
+    path: 26_headline_isfrontpage/train-00000-of-00001.parquet
+- config_name: 28_art_year
   data_files:
   - split: train
-    path: 9_nyc_long/train-*
-dataset_info:
-- config_name: 10_nyc_borough
-  features:
-  - name: prompt
-    dtype: string
-  - name: label
-    dtype: int32
-  - name: label_text
-    dtype: string
-  splits:
-  - name: train
-    num_bytes: 232329
-    num_examples: 5000
-  download_size: 103357
-  dataset_size: 232329
-- config_name: 11_us_lat
-  features:
-  - name: prompt
-    dtype: string
-  - name: label
-    dtype: int32
-  splits:
-  - name: train
-    num_bytes: 104949
-    num_examples: 5000
-  download_size: 70102
-  dataset_size: 104949
-- config_name: 12_us_long
-  features:
-  - name: prompt
-    dtype: string
-  - name: label
-    dtype: int32
-  splits:
-  - name: train
-    num_bytes: 105504
-    num_examples: 5000
-  download_size: 71809
-  dataset_size: 105504
-- config_name: 13_us_state
-  features:
-  - name: prompt
-    dtype: string
-  - name: label
-    dtype: int32
-  - name: label_text
-    dtype: string
-  splits:
-  - name: train
-    num_bytes: 136685
-    num_examples: 5000
-  download_size: 70779
-  dataset_size: 136685
-- config_name: 14_us_timezone
-  features:
-  - name: prompt
-    dtype: string
-  - name: label
-    dtype: int32
-  - name: label_text
-    dtype: string
-  splits:
-  - name: train
-    num_bytes: 169912
-    num_examples: 4998
-  download_size: 70601
-  dataset_size: 169912
-- config_name: 15_us_population
-  features:
-  - name: prompt
-    dtype: string
-  - name: label
-    dtype: int32
-  splits:
-  - name: train
-    num_bytes: 92463
-    num_examples: 5000
-  download_size: 83100
-  dataset_size: 92463
-- config_name: 16_us_density
-  features:
-  - name: prompt
-    dtype: string
-  - name: label
-    dtype: int32
-  splits:
-  - name: train
-    num_bytes: 92463
-    num_examples: 5000
-  download_size: 83100
-  dataset_size: 92463
-- config_name: 17_world_country
-  features:
-  - name: prompt
-    dtype: string
-  - name: label
-    dtype: int32
-  - name: label_text
-    dtype: string
-  splits:
-  - name: train
-    num_bytes: 189973
-    num_examples: 5000
-  download_size: 84315
-  dataset_size: 189973
-- config_name: 18_world_latitude
-  features:
-  - name: prompt
-    dtype: string
-  - name: label
-    dtype: int32
-  splits:
-  - name: train
-    num_bytes: 122212
-    num_examples: 5000
-  download_size: 89927
-  dataset_size: 122212
-- config_name: 19_world_longitude
-  features:
-  - name: prompt
-    dtype: string
-  - name: label
-    dtype: int32
-  splits:
-  - name: train
-    num_bytes: 121898
-    num_examples: 5000
-  download_size: 92552
-  dataset_size: 121898
-- config_name: 2_hist_fig_birthyear
-  features:
-  - name: prompt
-    dtype: string
-  - name: label
-    dtype: int32
-  splits:
-  - name: train
-    num_bytes: 134346
-    num_examples: 5000
-  download_size: 110070
-  dataset_size: 134346
-- config_name: 3_hist_fig_deathyear
-  features:
-  - name: prompt
-    dtype: string
-  - name: label
-    dtype: int32
-  splits:
-  - name: train
-    num_bytes: 134564
-    num_examples: 5000
-  download_size: 111719
-  dataset_size: 134564
-- config_name: 4_hist_fig_age
-  features:
-  - name: prompt
-    dtype: string
-  - name: label
-    dtype: int32
-  splits:
-  - name: train
-    num_bytes: 134535
-    num_examples: 5000
-  download_size: 97355
-  dataset_size: 134535
-- config_name: 5_hist_fig_ismale
-  features:
-  - name: prompt
-    dtype: string
-  - name: label
-    dtype: int32
-  splits:
-  - name: train
-    num_bytes: 141248
-    num_examples: 5000
-  download_size: 90870
-  dataset_size: 141248
-- config_name: 6_hist_fig_isamerican
-  features:
-  - name: prompt
-    dtype: string
-  - name: label
-    dtype: int32
-  splits:
-  - name: train
-    num_bytes: 124230
-    num_examples: 5000
-  download_size: 79499
-  dataset_size: 124230
-- config_name: 7_hist_fig_ispolitician
-  features:
-  - name: prompt
-    dtype: string
-  - name: label
-    dtype: int32
-  splits:
-  - name: train
-    num_bytes: 132799
-    num_examples: 5000
-  download_size: 88905
-  dataset_size: 132799
-- config_name: 8_nyc_lat
-  features:
-  - name: prompt
-    dtype: string
-  - name: label
-    dtype: int32
-  splits:
-  - name: train
-    num_bytes: 172220
-    num_examples: 5000
-  download_size: 99256
-  dataset_size: 172220
-- config_name: 9_nyc_long
-  features:
-  - name: prompt
-    dtype: string
-  - name: label
-    dtype: int32
-  splits:
-  - name: train
-    num_bytes: 171908
-    num_examples: 5000
-  download_size: 100403
-  dataset_size: 171908
+    path: 28_art_year/train-00000-of-00001.parquet
+- config_name: 29_art_pageviews
+  data_files:
+  - split: train
+    path: 29_art_pageviews/train-00000-of-00001.parquet
+- config_name: 30_book_length
+  data_files:
+  - split: train
+    path: 30_book_length/train-00000-of-00001.parquet
+- config_name: 36_sciq_tf
+  data_files:
+  - split: train
+    path: 36_sciq_tf/train-00000-of-00001.parquet
+- config_name: 39_arith_infix
+  data_files:
+  - split: train
+    path: 39_arith_infix/train-00000-of-00001.parquet
+- config_name: 40_arith_rpn
+  data_files:
+  - split: train
+    path: 40_arith_rpn/train-00000-of-00001.parquet
+- config_name: 44_phys_tf
+  data_files:
+  - split: train
+    path: 44_phys_tf/train-00000-of-00001.parquet
+- config_name: 45_context_ans
+  data_files:
+  - split: train
+    path: 45_context_ans/train-00000-of-00001.parquet
+- config_name: 48_cm_correct
+  data_files:
+  - split: train
+    path: 48_cm_correct/train-00000-of-00001.parquet
+- config_name: 49_cm_isshort
+  data_files:
+  - split: train
+    path: 49_cm_isshort/train-00000-of-00001.parquet
+- config_name: 50_deon_isvalid
+  data_files:
+  - split: train
+    path: 50_deon_isvalid/train-00000-of-00001.parquet
+- config_name: 51_just_is
+  data_files:
+  - split: train
+    path: 51_just_is/train-00000-of-00001.parquet
+- config_name: 52_virtue_is
+  data_files:
+  - split: train
+    path: 52_virtue_is/train-00000-of-00001.parquet
+- config_name: 57_wikidatais_alive
+  data_files:
+  - split: train
+    path: 57_wikidatais_alive/train-00000-of-00001.parquet
+- config_name: 59_wikidata_occupation_isjournalist
+  data_files:
+  - split: train
+    path: 59_wikidata_occupation_isjournalist/train-00000-of-00001.parquet
+- config_name: 60_wikidata_occupation_isathlete
+  data_files:
+  - split: train
+    path: 60_wikidata_occupation_isathlete/train-00000-of-00001.parquet
+- config_name: 61_wikidata_occupation_isactor
+  data_files:
+  - split: train
+    path: 61_wikidata_occupation_isactor/train-00000-of-00001.parquet
+- config_name: 62_wikidata_occupation_ispolitician
+  data_files:
+  - split: train
+    path: 62_wikidata_occupation_ispolitician/train-00000-of-00001.parquet
+- config_name: 63_wikidata_occupation_issinger
+  data_files:
+  - split: train
+    path: 63_wikidata_occupation_issinger/train-00000-of-00001.parquet
+- config_name: 64_wikidata_occupation_isresearcher
+  data_files:
+  - split: train
+    path: 64_wikidata_occupation_isresearcher/train-00000-of-00001.parquet
+- config_name: 86_social_iqa
+  data_files:
+  - split: train
+    path: 86_social_iqa/train-00000-of-00001.parquet
+- config_name: 87_glue_cola
+  data_files:
+  - split: train
+    path: 87_glue_cola/train-00000-of-00001.parquet
+- config_name: 89_glue_mrpc
+  data_files:
+  - split: train
+    path: 89_glue_mrpc/train-00000-of-00001.parquet
+- config_name: 90_glue_qnli
+  data_files:
+  - split: train
+    path: 90_glue_qnli/train-00000-of-00001.parquet
+- config_name: 91_glue_qqp
+  data_files:
+  - split: train
+    path: 91_glue_qqp/train-00000-of-00001.parquet
+- config_name: 92_glue_sst2
+  data_files:
+  - split: train
+    path: 92_glue_sst2/train-00000-of-00001.parquet
+- config_name: 93_glue_stsb
+  data_files:
+  - split: train
+    path: 93_glue_stsb/train-00000-of-00001.parquet
+- config_name: 94_ai_gen
+  data_files:
+  - split: train
+    path: 94_ai_gen/train-00000-of-00001.parquet
+- config_name: 95_toxic_is
+  data_files:
+  - split: train
+    path: 95_toxic_is/train-00000-of-00001.parquet
+- config_name: 96_spam_is
+  data_files:
+  - split: train
+    path: 96_spam_is/train-00000-of-00001.parquet
+- config_name: 100_news_fake
+  data_files:
+  - split: train
+    path: 100_news_fake/train-00000-of-00001.parquet
+- config_name: 106_hate_hate
+  data_files:
+  - split: train
+    path: 106_hate_hate/train-00000-of-00001.parquet
+- config_name: 107_hate_offensive
+  data_files:
+  - split: train
+    path: 107_hate_offensive/train-00000-of-00001.parquet
+- config_name: 109_ball_ws
+  data_files:
+  - split: train
+    path: 109_ball_ws/train-00000-of-00001.parquet
+- config_name: 110_aimade_humangpt3
+  data_files:
+  - split: train
+    path: 110_aimade_humangpt3/train-00000-of-00001.parquet
+- config_name: 111_yelp_data
+  data_files:
+  - split: train
+    path: 111_yelp_data/train-00000-of-00001.parquet
+- config_name: 112_amzn_rev
+  data_files:
+  - split: train
+    path: 112_amzn_rev/train-00000-of-00001.parquet
+- config_name: 113_movie_sent
+  data_files:
+  - split: train
+    path: 113_movie_sent/train-00000-of-00001.parquet
+- config_name: 114_nyc_borough_Manhattan
+  data_files:
+  - split: train
+    path: 114_nyc_borough_Manhattan/train-00000-of-00001.parquet
+- config_name: 115_nyc_borough_Brooklyn
+  data_files:
+  - split: train
+    path: 115_nyc_borough_Brooklyn/train-00000-of-00001.parquet
+- config_name: 116_nyc_borough_Bronx
+  data_files:
+  - split: train
+    path: 116_nyc_borough_Bronx/train-00000-of-00001.parquet
+- config_name: 117_us_state_FL
+  data_files:
+  - split: train
+    path: 117_us_state_FL/train-00000-of-00001.parquet
+- config_name: 118_us_state_CA
+  data_files:
+  - split: train
+    path: 118_us_state_CA/train-00000-of-00001.parquet
+- config_name: 119_us_state_TX
+  data_files:
+  - split: train
+    path: 119_us_state_TX/train-00000-of-00001.parquet
+- config_name: 120_us_timezone_Chicago
+  data_files:
+  - split: train
+    path: 120_us_timezone_Chicago/train-00000-of-00001.parquet
+- config_name: 121_us_timezone_New_York
+  data_files:
+  - split: train
+    path: 121_us_timezone_New_York/train-00000-of-00001.parquet
+- config_name: 122_us_timezone_Los_Angeles
+  data_files:
+  - split: train
+    path: 122_us_timezone_Los_Angeles/train-00000-of-00001.parquet
+- config_name: 123_world_country_United_Kingdom
+  data_files:
+  - split: train
+    path: 123_world_country_United_Kingdom/train-00000-of-00001.parquet
+- config_name: 124_world_country_United_States
+  data_files:
+  - split: train
+    path: 124_world_country_United_States/train-00000-of-00001.parquet
+- config_name: 125_world_country_Italy
+  data_files:
+  - split: train
+    path: 125_world_country_Italy/train-00000-of-00001.parquet
+- config_name: 126_art_type_book
+  data_files:
+  - split: train
+    path: 126_art_type_book/train-00000-of-00001.parquet
+- config_name: 127_art_type_song
+  data_files:
+  - split: train
+    path: 127_art_type_song/train-00000-of-00001.parquet
+- config_name: 128_art_type_movie
+  data_files:
+  - split: train
+    path: 128_art_type_movie/train-00000-of-00001.parquet
+- config_name: 129_arith_mc_A
+  data_files:
+  - split: train
+    path: 129_arith_mc_A/train-00000-of-00001.parquet
+- config_name: 130_temp_cat_Frequency
+  data_files:
+  - split: train
+    path: 130_temp_cat_Frequency/train-00000-of-00001.parquet
+- config_name: 131_temp_cat_Typical Time
+  data_files:
+  - split: train
+    path: 131_temp_cat_Typical Time/train-00000-of-00001.parquet
+- config_name: 132_temp_cat_Event Ordering
+  data_files:
+  - split: train
+    path: 132_temp_cat_Event Ordering/train-00000-of-00001.parquet
+- config_name: 133_context_type_Causality
+  data_files:
+  - split: train
+    path: 133_context_type_Causality/train-00000-of-00001.parquet
+- config_name: 134_context_type_Belief_states
+  data_files:
+  - split: train
+    path: 134_context_type_Belief_states/train-00000-of-00001.parquet
+- config_name: 135_context_type_Event_duration
+  data_files:
+  - split: train
+    path: 135_context_type_Event_duration/train-00000-of-00001.parquet
+- config_name: 136_glue_mnli_entailment
+  data_files:
+  - split: train
+    path: 136_glue_mnli_entailment/train-00000-of-00001.parquet
+- config_name: 137_glue_mnli_neutral
+  data_files:
+  - split: train
+    path: 137_glue_mnli_neutral/train-00000-of-00001.parquet
+- config_name: 138_glue_mnli_contradiction
+  data_files:
+  - split: train
+    path: 138_glue_mnli_contradiction/train-00000-of-00001.parquet
+- config_name: 139_news_class_Politics
+  data_files:
+  - split: train
+    path: 139_news_class_Politics/train-00000-of-00001.parquet
+- config_name: 140_news_class_Technology
+  data_files:
+  - split: train
+    path: 140_news_class_Technology/train-00000-of-00001.parquet
+- config_name: 141_news_class_Entertainment
+  data_files:
+  - split: train
+    path: 141_news_class_Entertainment/train-00000-of-00001.parquet
+- config_name: 142_cancer_cat_Thyroid_Cancer
+  data_files:
+  - split: train
+    path: 142_cancer_cat_Thyroid_Cancer/train-00000-of-00001.parquet
+- config_name: 143_cancer_cat_Lung_Cancer
+  data_files:
+  - split: train
+    path: 143_cancer_cat_Lung_Cancer/train-00000-of-00001.parquet
+- config_name: 144_cancer_cat_Colon_Cancer
+  data_files:
+  - split: train
+    path: 144_cancer_cat_Colon_Cancer/train-00000-of-00001.parquet
+- config_name: 145_disease_class_digestive system diseases
+  data_files:
+  - split: train
+    path: 145_disease_class_digestive system diseases/train-00000-of-00001.parquet
+- config_name: 146_disease_class_cardiovascular diseases
+  data_files:
+  - split: train
+    path: 146_disease_class_cardiovascular diseases/train-00000-of-00001.parquet
+- config_name: 147_disease_class_nervous system diseases
+  data_files:
+  - split: train
+    path: 147_disease_class_nervous system diseases/train-00000-of-00001.parquet
+- config_name: 148_twt_emotion_worry
+  data_files:
+  - split: train
+    path: 148_twt_emotion_worry/train-00000-of-00001.parquet
+- config_name: 149_twt_emotion_happiness
+  data_files:
+  - split: train
+    path: 149_twt_emotion_happiness/train-00000-of-00001.parquet
+- config_name: 150_twt_emotion_sadness
+  data_files:
+  - split: train
+    path: 150_twt_emotion_sadness/train-00000-of-00001.parquet
+- config_name: 151_it_tick_HR Support
+  data_files:
+  - split: train
+    path: 151_it_tick_HR Support/train-00000-of-00001.parquet
+- config_name: 152_it_tick_Hardware
+  data_files:
+  - split: train
+    path: 152_it_tick_Hardware/train-00000-of-00001.parquet
+- config_name: 153_it_tick_Administrative rights
+  data_files:
+  - split: train
+    path: 153_it_tick_Administrative rights/train-00000-of-00001.parquet
+- config_name: 154_athlete_sport_football
+  data_files:
+  - split: train
+    path: 154_athlete_sport_football/train-00000-of-00001.parquet
+- config_name: 155_athlete_sport_basketball
+  data_files:
+  - split: train
+    path: 155_athlete_sport_basketball/train-00000-of-00001.parquet
+- config_name: 156_athlete_sport_baseball
+  data_files:
+  - split: train
+    path: 156_athlete_sport_baseball/train-00000-of-00001.parquet
+- config_name: 157_amazon_5star
+  data_files:
+  - split: train
+    path: 157_amazon_5star/train-00000-of-00001.parquet
+- config_name: 158_code_C
+  data_files:
+  - split: train
+    path: 158_code_C/train-00000-of-00001.parquet
+- config_name: 159_code_Python
+  data_files:
+  - split: train
+    path: 159_code_Python/train-00000-of-00001.parquet
+- config_name: 160_code_HTML
+  data_files:
+  - split: train
+    path: 160_code_HTML/train-00000-of-00001.parquet
+- config_name: 161_agnews_0
+  data_files:
+  - split: train
+    path: 161_agnews_0/train-00000-of-00001.parquet
+- config_name: 162_agnews_1
+  data_files:
+  - split: train
+    path: 162_agnews_1/train-00000-of-00001.parquet
+- config_name: 163_agnews_2
+  data_files:
+  - split: train
+    path: 163_agnews_2/train-00000-of-00001.parquet
 ---
 
 # Sparse Probing Datasets
 
-155 binary classification datasets for probing LLM representations.
+**113 binary classification tasks** for probing language model representations.
 
-From: ["Are Sparse Autoencoders Useful? A Case Study in Sparse Probing"](https://arxiv.org/abs/2502.16681)
+From: "Are Sparse Autoencoders Useful? A Case Study in Sparse Probing" ([arXiv:2502.16681](https://arxiv.org/abs/2502.16681))
+
+Source: [EleutherAI/sae-probes](https://github.com/EleutherAI/sae-probes)
 
 ## Usage
 
@@ -325,180 +470,197 @@ from datasets import load_dataset
 # Load a specific dataset
 ds = load_dataset("serteal/sparse-probing", "87_glue_cola")
 
-# Access the data
-for example in ds["train"]:
-    print(example["prompt"], example["label"])
+# List available configurations
+from datasets import get_dataset_config_names
+configs = get_dataset_config_names("serteal/sparse-probing")
+print(f"Available: {len(configs)} datasets")
 ```
 
 ## Dataset Categories
 
 | Category | Count | Description |
 |----------|-------|-------------|
-| knowledge | 42 | Historical figures, geography, wikidata |
-| glue | 10 | GLUE benchmark tasks |
-| reasoning | 27 | TruthfulQA, ethics, commonsense |
-| moderation | 16 | Toxicity, spam, fake news |
-| sentiment | 9 | Reviews, emotions |
-| science | 27 | Domain concepts, medical |
+| knowledge | 50 | Historical figures, geography, wikidata |
+| reasoning | 18 | Science, ethics, commonsense, temporal |
+| science | 6 | Domain concepts, medical, biology |
+| moderation | 13 | Toxicity, spam, fake news, hate speech |
+| glue | 9 | GLUE benchmark tasks |
+| sentiment | 7 | Emotion, reviews |
+| other | 7 | IT tickets, sports |
 | code | 3 | Programming language detection |
-| other | 21 | IT tickets, sports |
-
-## Available Configurations
-
-| ID | Name | Category | Description |
-|----|------|----------|-------------|
-| 2 | hist_fig_birthyear | knowledge | Birth year of historical figures |
-| 3 | hist_fig_deathyear | knowledge | Death year of historical figures |
-| 4 | hist_fig_age | knowledge | Age of historical figures |
-| 5 | hist_fig_ismale | knowledge | Gender classification of historical figures |
-| 6 | hist_fig_isamerican | knowledge | American nationality of historical figures |
-| 7 | hist_fig_ispolitician | knowledge | Politician occupation of historical figures |
-| 8 | nyc_lat | knowledge | NYC latitude classification |
-| 9 | nyc_long | knowledge | NYC longitude classification |
-| 10 | nyc_borough | knowledge | NYC borough classification |
-| 11 | us_lat | knowledge | US latitude classification |
-| 12 | us_long | knowledge | US longitude classification |
-| 13 | us_state | knowledge | US state classification |
-| 14 | us_timezone | knowledge | US timezone classification |
-| 15 | us_population | knowledge | US population classification |
-| 16 | us_density | knowledge | US density classification |
-| 17 | world_country | knowledge | World country classification |
-| 18 | world_latitude | knowledge | World latitude classification |
-| 19 | world_longitude | knowledge | World longitude classification |
-| 20 | world_pageviews | knowledge | World entity pageviews classification |
-| 21 | headline_istrump | knowledge | Headlines mentioning Trump |
-| 22 | headline_isobama | knowledge | Headlines mentioning Obama |
-| 23 | headline_ischina | knowledge | Headlines mentioning China |
-| 24 | headline_isiran | knowledge | Headlines mentioning Iran |
-| 25 | headline_year | knowledge | Headline year classification |
-| 26 | headline_isfrontpage | knowledge | Front page headline classification |
-| 27 | art_type | knowledge | Art type classification (book/song/movie) |
-| 28 | art_year | knowledge | Art year classification |
-| 29 | art_pageviews | knowledge | Art pageviews classification |
-| 30 | book_length | knowledge | Book length classification |
-| 36 | sciq_tf | reasoning | Science questions true/false |
-| 37 | arith_mc | reasoning | Arithmetic multiple choice |
-| 38 | arith_difficulty | reasoning | Arithmetic difficulty classification |
-| 39 | arith_infix | reasoning | Arithmetic infix notation |
-| 40 | arith_rpn | reasoning | Arithmetic reverse polish notation |
-| 41 | truthqa_tf | reasoning | TruthfulQA true/false |
-| 42 | temp_sense | reasoning | Temporal sense classification |
-| 43 | temp_cat | reasoning | Temporal category classification |
-| 44 | phys_tf | reasoning | Physical reasoning true/false |
-| 45 | context_ans | reasoning | Contextual answer classification |
-| 46 | context_type | reasoning | Context type classification |
-| 47 | reasoning_tf | reasoning | Commonsense reasoning true/false |
-| 48 | cm_correct | reasoning | Common morality correctness |
-| 49 | cm_isshort | reasoning | Common morality statement brevity |
-| 50 | deon_isvalid | reasoning | Deontological validity classification |
-| 51 | just_is | reasoning | Justice classification |
-| 52 | virtue_is | reasoning | Virtue ethics classification |
-| 54 | cs_tf | reasoning | Commonsense true/false |
-| 55 | open_qa | reasoning | Open-domain QA classification |
-| 56 | wikidatasex_or_gender | knowledge | Wikidata gender classification |
-| 57 | wikidatais_alive | knowledge | Wikidata alive status |
-| 58 | wikidatapolitical_party | knowledge | Wikidata political party |
-| 59 | wikidata_occupation_isjournalist | knowledge | Wikidata journalist occupation |
-| 60 | wikidata_occupation_isathlete | knowledge | Wikidata athlete occupation |
-| 61 | wikidata_occupation_isactor | knowledge | Wikidata actor occupation |
-| 62 | wikidata_occupation_ispolitician | knowledge | Wikidata politician occupation |
-| 63 | wikidata_occupation_issinger | knowledge | Wikidata singer occupation |
-| 64 | wikidata_occupation_isresearcher | knowledge | Wikidata researcher occupation |
-| 65 | high-school | science | High school concept recognition |
-| 66 | living-room | science | Living room concept recognition |
-| 67 | social-security | science | Social security concept recognition |
-| 68 | credit-card | science | Credit card concept recognition |
-| 69 | blood-pressure | science | Blood pressure concept recognition |
-| 70 | prime-factors | science | Prime factors concept recognition |
-| 71 | social-media | science | Social media concept recognition |
-| 72 | gene-expression | science | Gene expression concept recognition |
-| 73 | control-group | science | Control group concept recognition |
-| 74 | magnetic-field | science | Magnetic field concept recognition |
-| 75 | cell-lines | science | Cell lines concept recognition |
-| 76 | trial-court | science | Trial court concept recognition |
-| 77 | second-derivative | science | Second derivative concept recognition |
-| 78 | north-america | science | North America concept recognition |
-| 79 | human-rights | science | Human rights concept recognition |
-| 80 | side-effects | science | Side effects concept recognition |
-| 81 | public-health | science | Public health concept recognition |
-| 82 | federal-government | science | Federal government concept recognition |
-| 83 | third-party | science | Third party concept recognition |
-| 84 | clinical-trials | science | Clinical trials concept recognition |
-| 85 | mental-health | science | Mental health concept recognition |
-| 86 | social_iqa | reasoning | Social intelligence QA |
-| 87 | glue_cola | glue | GLUE CoLA linguistic acceptability |
-| 88 | glue_mnli | glue | GLUE MNLI natural language inference |
-| 89 | glue_mrpc | glue | GLUE MRPC paraphrase detection |
-| 90 | glue_qnli | glue | GLUE QNLI question answering NLI |
-| 91 | glue_qqp | glue | GLUE QQP Quora question pairs |
-| 92 | glue_sst2 | glue | GLUE SST2 sentiment analysis |
-| 93 | glue_stsb | glue | GLUE STSB semantic similarity |
-| 94 | ai_gen | moderation | AI-generated text detection |
-| 95 | toxic_is | moderation | Toxicity detection |
-| 96 | spam_is | moderation | Spam detection |
-| 97 | news_class | moderation | News category classification |
-| 98 | cancer_cat | science | Cancer category classification |
-| 99 | sci_cat | science | Science category classification |
-| 100 | news_fake | moderation | Fake news detection |
-| 101 | disease_class | science | Disease classification |
-| 102 | twt_emotion | sentiment | Tweet emotion classification |
-| 103 | it_tick | other | IT ticket classification |
-| 105 | click_bait | moderation | Clickbait detection |
-| 106 | hate_hate | moderation | Hate speech detection |
-| 107 | hate_offensive | moderation | Offensive language detection |
-| 108 | athlete_sport | other | Athlete sport classification |
-| 109 | ball_ws | other | Basketball win shares classification |
-| 110 | aimade_humangpt3 | moderation | Human vs GPT-3 text detection |
-| 111 | yelp_data | sentiment | Yelp review sentiment |
-| 112 | amzn_rev | sentiment | Amazon review sentiment |
-| 113 | movie_sent | sentiment | Movie review sentiment |
-| 114 | nyc_borough_Manhattan | knowledge | NYC Manhattan borough |
-| 115 | nyc_borough_Brooklyn | knowledge | NYC Brooklyn borough |
-| 116 | nyc_borough_Bronx | knowledge | NYC Bronx borough |
-| 117 | us_state_FL | knowledge | US Florida state |
-| 118 | us_state_CA | knowledge | US California state |
-| 119 | us_state_TX | knowledge | US Texas state |
-| 120 | us_timezone_Chicago | knowledge | US Chicago timezone |
-| 121 | us_timezone_New_York | knowledge | US New York timezone |
-| 122 | us_timezone_Los_Angeles | knowledge | US Los Angeles timezone |
-| 123 | world_country_United_Kingdom | knowledge | World United Kingdom |
-| 124 | world_country_United_States | knowledge | World United States |
-| 125 | world_country_Italy | knowledge | World Italy |
-| 126 | art_type_book | knowledge | Art type book classification |
-| 127 | art_type_song | knowledge | Art type song classification |
-| 128 | art_type_movie | knowledge | Art type movie classification |
-| 129 | arith_mc_A | reasoning | Arithmetic multiple choice A |
-| 130 | temp_cat_Frequency | reasoning | Temporal category frequency |
-| 131 | temp_cat_Typical Time | reasoning | Temporal category typical time |
-| 132 | temp_cat_Event Ordering | reasoning | Temporal category event ordering |
-| 133 | context_type_Causality | reasoning | Context type causality |
-| 134 | context_type_Belief_states | reasoning | Context type belief states |
-| 135 | context_type_Event_duration | reasoning | Context type event duration |
-| 136 | glue_mnli_entailment | glue | GLUE MNLI entailment |
-| 137 | glue_mnli_neutral | glue | GLUE MNLI neutral |
-| 138 | glue_mnli_contradiction | glue | GLUE MNLI contradiction |
-| 139 | news_class_Politics | moderation | News politics classification |
-| 140 | news_class_Technology | moderation | News technology classification |
-| 141 | news_class_Entertainment | moderation | News entertainment classification |
-| 142 | cancer_cat_Thyroid_Cancer | science | Thyroid cancer classification |
-| 143 | cancer_cat_Lung_Cancer | science | Lung cancer classification |
-| 144 | cancer_cat_Colon_Cancer | science | Colon cancer classification |
-| 145 | disease_class_digestive system diseases | science | Digestive system diseases |
-| 146 | disease_class_cardiovascular diseases | science | Cardiovascular diseases |
-| 147 | disease_class_nervous system diseases | science | Nervous system diseases |
-| 148 | twt_emotion_worry | sentiment | Tweet emotion worry |
-| 149 | twt_emotion_happiness | sentiment | Tweet emotion happiness |
-| 150 | twt_emotion_sadness | sentiment | Tweet emotion sadness |
-| 151 | it_tick_HR Support | other | IT ticket HR support |
-| 152 | it_tick_Hardware | other | IT ticket hardware |
-| 153 | it_tick_Administrative rights | other | IT ticket administrative rights |
-| 154 | athlete_sport_football | other | Athlete football classification |
-| 155 | athlete_sport_basketball | other | Athlete basketball classification |
-| 156 | athlete_sport_baseball | other | Athlete baseball classification |
-| 157 | amazon_5star | sentiment | Amazon 5-star rating classification |
-| 158 | code_C | code | C code detection |
-| 159 | code_Python | code | Python code detection |
-| 160 | code_HTML | code | HTML code detection |
-| 161 | agnews_0 | moderation | AG News category 0 |
-| 162 | agnews_1 | moderation | AG News category 1 |
-| 163 | agnews_2 | moderation | AG News category 2 |
+
+## Available Datasets
+
+### Knowledge
+
+| Config Name | Description |
+|-------------|-------------|
+| `2_hist_fig_birthyear` | Birth year of historical figures |
+| `3_hist_fig_deathyear` | Death year of historical figures |
+| `4_hist_fig_age` | Age of historical figures |
+| `5_hist_fig_ismale` | Gender classification of historical figures |
+| `6_hist_fig_isamerican` | American nationality of historical figures |
+| `7_hist_fig_ispolitician` | Politician occupation of historical figures |
+| `8_nyc_lat` | NYC latitude classification |
+| `9_nyc_long` | NYC longitude classification |
+| `10_nyc_borough` | NYC borough classification |
+| `11_us_lat` | US latitude classification |
+| `12_us_long` | US longitude classification |
+| `13_us_state` | US state classification |
+| `14_us_timezone` | US timezone classification |
+| `15_us_population` | US population classification |
+| `16_us_density` | US density classification |
+| `17_world_country` | World country classification |
+| `18_world_latitude` | World latitude classification |
+| `19_world_longitude` | World longitude classification |
+| `20_world_pageviews` | World entity pageviews classification |
+| `21_headline_istrump` | Headlines mentioning Trump |
+| `22_headline_isobama` | Headlines mentioning Obama |
+| `23_headline_ischina` | Headlines mentioning China |
+| `24_headline_isiran` | Headlines mentioning Iran |
+| `25_headline_year` | Headline year classification |
+| `26_headline_isfrontpage` | Front page headline classification |
+| `28_art_year` | Art year classification |
+| `29_art_pageviews` | Art pageviews classification |
+| `30_book_length` | Book length classification |
+| `57_wikidatais_alive` | Wikidata alive status |
+| `59_wikidata_occupation_isjournalist` | Wikidata journalist occupation |
+| `60_wikidata_occupation_isathlete` | Wikidata athlete occupation |
+| `61_wikidata_occupation_isactor` | Wikidata actor occupation |
+| `62_wikidata_occupation_ispolitician` | Wikidata politician occupation |
+| `63_wikidata_occupation_issinger` | Wikidata singer occupation |
+| `64_wikidata_occupation_isresearcher` | Wikidata researcher occupation |
+| `114_nyc_borough_Manhattan` | NYC Manhattan borough |
+| `115_nyc_borough_Brooklyn` | NYC Brooklyn borough |
+| `116_nyc_borough_Bronx` | NYC Bronx borough |
+| `117_us_state_FL` | US Florida state |
+| `118_us_state_CA` | US California state |
+| `119_us_state_TX` | US Texas state |
+| `120_us_timezone_Chicago` | US Chicago timezone |
+| `121_us_timezone_New_York` | US New York timezone |
+| `122_us_timezone_Los_Angeles` | US Los Angeles timezone |
+| `123_world_country_United_Kingdom` | World United Kingdom |
+| `124_world_country_United_States` | World United States |
+| `125_world_country_Italy` | World Italy |
+| `126_art_type_book` | Art type book classification |
+| `127_art_type_song` | Art type song classification |
+| `128_art_type_movie` | Art type movie classification |
+
+### Reasoning
+
+| Config Name | Description |
+|-------------|-------------|
+| `36_sciq_tf` | Science questions true/false |
+| `39_arith_infix` | Arithmetic infix notation |
+| `40_arith_rpn` | Arithmetic reverse polish notation |
+| `44_phys_tf` | Physical reasoning true/false |
+| `45_context_ans` | Contextual answer classification |
+| `48_cm_correct` | Common morality correctness |
+| `49_cm_isshort` | Common morality statement brevity |
+| `50_deon_isvalid` | Deontological validity classification |
+| `51_just_is` | Justice classification |
+| `52_virtue_is` | Virtue ethics classification |
+| `86_social_iqa` | Social intelligence QA |
+| `129_arith_mc_A` | Arithmetic multiple choice A |
+| `130_temp_cat_Frequency` | Temporal category frequency |
+| `131_temp_cat_Typical Time` | Temporal category typical time |
+| `132_temp_cat_Event Ordering` | Temporal category event ordering |
+| `133_context_type_Causality` | Context type causality |
+| `134_context_type_Belief_states` | Context type belief states |
+| `135_context_type_Event_duration` | Context type event duration |
+
+### Science
+
+| Config Name | Description |
+|-------------|-------------|
+| `142_cancer_cat_Thyroid_Cancer` | Thyroid cancer classification |
+| `143_cancer_cat_Lung_Cancer` | Lung cancer classification |
+| `144_cancer_cat_Colon_Cancer` | Colon cancer classification |
+| `145_disease_class_digestive system diseases` | Digestive system diseases |
+| `146_disease_class_cardiovascular diseases` | Cardiovascular diseases |
+| `147_disease_class_nervous system diseases` | Nervous system diseases |
+
+### Moderation
+
+| Config Name | Description |
+|-------------|-------------|
+| `94_ai_gen` | AI-generated text detection |
+| `95_toxic_is` | Toxicity detection |
+| `96_spam_is` | Spam detection |
+| `100_news_fake` | Fake news detection |
+| `106_hate_hate` | Hate speech detection |
+| `107_hate_offensive` | Offensive language detection |
+| `110_aimade_humangpt3` | Human vs GPT-3 text detection |
+| `139_news_class_Politics` | News politics classification |
+| `140_news_class_Technology` | News technology classification |
+| `141_news_class_Entertainment` | News entertainment classification |
+| `161_agnews_0` | AG News category 0 |
+| `162_agnews_1` | AG News category 1 |
+| `163_agnews_2` | AG News category 2 |
+
+### Glue
+
+| Config Name | Description |
+|-------------|-------------|
+| `87_glue_cola` | GLUE CoLA linguistic acceptability |
+| `89_glue_mrpc` | GLUE MRPC paraphrase detection |
+| `90_glue_qnli` | GLUE QNLI question answering NLI |
+| `91_glue_qqp` | GLUE QQP Quora question pairs |
+| `92_glue_sst2` | GLUE SST2 sentiment analysis |
+| `93_glue_stsb` | GLUE STSB semantic similarity |
+| `136_glue_mnli_entailment` | GLUE MNLI entailment |
+| `137_glue_mnli_neutral` | GLUE MNLI neutral |
+| `138_glue_mnli_contradiction` | GLUE MNLI contradiction |
+
+### Sentiment
+
+| Config Name | Description |
+|-------------|-------------|
+| `111_yelp_data` | Yelp review sentiment |
+| `112_amzn_rev` | Amazon review sentiment |
+| `113_movie_sent` | Movie review sentiment |
+| `148_twt_emotion_worry` | Tweet emotion worry |
+| `149_twt_emotion_happiness` | Tweet emotion happiness |
+| `150_twt_emotion_sadness` | Tweet emotion sadness |
+| `157_amazon_5star` | Amazon 5-star rating classification |
+
+### Other
+
+| Config Name | Description |
+|-------------|-------------|
+| `109_ball_ws` | Basketball win shares classification |
+| `151_it_tick_HR Support` | IT ticket HR support |
+| `152_it_tick_Hardware` | IT ticket hardware |
+| `153_it_tick_Administrative rights` | IT ticket administrative rights |
+| `154_athlete_sport_football` | Athlete football classification |
+| `155_athlete_sport_basketball` | Athlete basketball classification |
+| `156_athlete_sport_baseball` | Athlete baseball classification |
+
+### Code
+
+| Config Name | Description |
+|-------------|-------------|
+| `158_code_C` | C code detection |
+| `159_code_Python` | Python code detection |
+| `160_code_HTML` | HTML code detection |
+
+## Schema
+
+Each dataset contains:
+- `prompt` (string): The input text
+- `label` (int): Binary label (0 or 1)
+
+## Citation
+
+```bibtex
+@article{makelov2025sparse,
+  title={Are Sparse Autoencoders Useful? A Case Study in Sparse Probing},
+  author={Makelov, Aleksandar and others},
+  journal={arXiv preprint arXiv:2502.16681},
+  year={2025}
+}
+```
+
+## License
+
+Apache 2.0