meg HF staff commited on
Commit
2430327
2 Parent(s): b94303d 900f15c

Merge branch 'main' of https://huggingface.co/spaces/huggingface/data-measurements-tool-2 into main

Browse files
This view is limited to 50 files because it contains too many changes.   See raw diff
Files changed (50) hide show
  1. app.py +1 -1
  2. cache_dir/{bak/zipf_basic_stats.json → c4_en.noblocklist_train_text/node_figure.json} +2 -2
  3. cache_dir/{bak/zipf_fig.json → c4_en.noblocklist_train_text/node_list.th} +2 -2
  4. cache_dir/c4_en_train_text/node_figure.json +3 -0
  5. cache_dir/c4_en_train_text/node_list.th +2 -2
  6. cache_dir/c4_realnewslike_train_text/node_figure.json +3 -0
  7. cache_dir/c4_realnewslike_train_text/node_list.th +3 -0
  8. cache_dir/glue_mnli_train_hypothesis/node_figure.json +3 -0
  9. cache_dir/glue_mnli_train_hypothesis/node_list.th +2 -2
  10. cache_dir/glue_mrpc_validation_sentence1/node_list.th +3 -0
  11. cache_dir/glue_qqp_train_question1/node_figure.json +3 -0
  12. cache_dir/glue_qqp_train_question1/node_list.th +2 -2
  13. cache_dir/glue_rte_validation_sentence1/node_list.th +3 -0
  14. cache_dir/glue_rte_validation_sentence2/node_list.th +3 -0
  15. cache_dir/glue_stsb_train_sentence2/node_figure.json +3 -0
  16. cache_dir/glue_stsb_train_sentence2/node_list.th +3 -0
  17. cache_dir/glue_stsb_validation_sentence1/node_figure.json +3 -0
  18. cache_dir/glue_stsb_validation_sentence1/node_list.th +3 -0
  19. cache_dir/glue_stsb_validation_sentence2/node_figure.json +3 -0
  20. cache_dir/glue_stsb_validation_sentence2/node_list.th +3 -0
  21. cache_dir/glue_wnli_validation_sentence1/node_list.th +3 -0
  22. cache_dir/glue_wnli_validation_sentence2/node_list.th +3 -0
  23. cache_dir/hate_speech18_default_train_text/fig_tok_length.png +3 -0
  24. cache_dir/hate_speech18_default_train_text/node_list.th +3 -0
  25. cache_dir/hate_speech_offensive_default_train_tweet/node_list.th +3 -0
  26. cache_dir/squad_plain_text_train_context/node_figure.json +3 -0
  27. cache_dir/squad_plain_text_train_context/node_list.th +3 -0
  28. cache_dir/squad_plain_text_train_question/node_figure.json +3 -0
  29. cache_dir/squad_plain_text_train_question/node_list.th +3 -0
  30. cache_dir/squad_plain_text_train_title/node_figure.json +3 -0
  31. cache_dir/squad_plain_text_train_title/node_list.th +3 -0
  32. cache_dir/squad_plain_text_validation_context/node_figure.json +3 -0
  33. cache_dir/squad_plain_text_validation_context/node_list.th +3 -0
  34. cache_dir/squad_plain_text_validation_question/node_figure.json +3 -0
  35. cache_dir/squad_plain_text_validation_question/node_list.th +3 -0
  36. cache_dir/squad_plain_text_validation_title/node_figure.json +3 -0
  37. cache_dir/squad_plain_text_validation_title/node_list.th +3 -0
  38. cache_dir/squad_v2_squad_v2_train_context/node_figure.json +3 -0
  39. cache_dir/squad_v2_squad_v2_train_context/node_list.th +3 -0
  40. cache_dir/squad_v2_squad_v2_train_question/node_figure.json +3 -0
  41. cache_dir/squad_v2_squad_v2_train_question/node_list.th +3 -0
  42. cache_dir/squad_v2_squad_v2_train_title/node_figure.json +3 -0
  43. cache_dir/squad_v2_squad_v2_train_title/node_list.th +3 -0
  44. cache_dir/squad_v2_squad_v2_validation_context/node_figure.json +3 -0
  45. cache_dir/squad_v2_squad_v2_validation_context/node_list.th +3 -0
  46. cache_dir/squad_v2_squad_v2_validation_question/node_figure.json +3 -0
  47. cache_dir/squad_v2_squad_v2_validation_question/node_list.th +3 -0
  48. cache_dir/squad_v2_squad_v2_validation_title/node_figure.json +3 -0
  49. cache_dir/squad_v2_squad_v2_validation_title/node_list.th +3 -0
  50. cache_dir/super_glue_boolq_train_passage/node_figure.json +3 -0
app.py CHANGED
@@ -216,7 +216,7 @@ def main():
216
 
217
  # When not doing new development, use the cache.
218
  use_cache = True
219
- show_embeddings = st.sidebar.checkbox("Show embeddings")
220
  # List of datasets for which embeddings are hard to compute:
221
 
222
  if compare_mode:
 
216
 
217
  # When not doing new development, use the cache.
218
  use_cache = True
219
+ show_embeddings = st.sidebar.checkbox("Show text clusters")
220
  # List of datasets for which embeddings are hard to compute:
221
 
222
  if compare_mode:
cache_dir/{bak/zipf_basic_stats.json → c4_en.noblocklist_train_text/node_figure.json} RENAMED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:a98e33063e6a463e1fd5ba712159e0d665a13c051455736ac61ec13635cbefb9
3
- size 62843
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:d747047909d1c71dfab01b488bde25ba652d8e40a2f23b15a15e45f53f8359d7
3
+ size 93271
cache_dir/{bak/zipf_fig.json → c4_en.noblocklist_train_text/node_list.th} RENAMED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:e05f628ddd584cc8bef475935053754bd6878e74b08229486f45378fb3a6003b
3
- size 15959382
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:aab25e321004a413bacfeef6125e595dcf1ccd7e957bbc41efa29524c04f4bef
3
+ size 23692149
cache_dir/c4_en_train_text/node_figure.json ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:59873fee6b131784687730bbaf8f54b7537a5781088d54211dc39ea8f868157d
3
+ size 91467
cache_dir/c4_en_train_text/node_list.th CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:e1a21a6fd19ae3b95c281d1325eae02a0e64e1a75bac7b01aca759245b209b5c
3
- size 24724129
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:044b98e15c690ce7eba2a3ddcb9e721cbb80905ff4a6dadb9026e26e6e24bf44
3
+ size 23525761
cache_dir/c4_realnewslike_train_text/node_figure.json ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:5d03cea1f2728c912e3fc3649c85efc589115d1fd460cd672e3272ee71f906a4
3
+ size 110983
cache_dir/c4_realnewslike_train_text/node_list.th ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:459d5ba497e165d0140946f468526da838cb3d8ce800b8dcb396f39e6231af30
3
+ size 29335085
cache_dir/glue_mnli_train_hypothesis/node_figure.json ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:01e77deba87cf5279f9f9a838293abf7e9e7718d8553023b4441bfb27d6244fb
3
+ size 70067
cache_dir/glue_mnli_train_hypothesis/node_list.th CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:927e878778df79dcd232a8f2a3b22eb92548a661104b790699076bbc9bff7a33
3
- size 25562251
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:31e13c4cccfbb43d4727f09db09085e950381cc208205b24b6347c41ecc264b9
3
+ size 23985089
cache_dir/glue_mrpc_validation_sentence1/node_list.th ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:dcaac341943444031cbd7271dfdd4bccc5b51fea8e0df8584f1522b3722e39ee
3
+ size 135885
cache_dir/glue_qqp_train_question1/node_figure.json ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:8665fd2eba8166719c06c56f8cc57e6abb806806187a658766f27e4d6b461a5b
3
+ size 512444
cache_dir/glue_qqp_train_question1/node_list.th CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:6a5c2314ea944bdfd8307ee0374e986cad385806977c9e7ef3a413022c5406ee
3
- size 44583251
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:12a301100bed49b2b7a81d1bbe7c8b79d4bd0b7429c90b62e7a7ec3f4ff01ae1
3
+ size 36825041
cache_dir/glue_rte_validation_sentence1/node_list.th ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:321b1bfb25918814eee8ee026aaa3185075390eca900c623231520d4a5d9b6a0
3
+ size 101699
cache_dir/glue_rte_validation_sentence2/node_list.th ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:754eb5720317280e325b9fff1f6e8f67181e0bb713feed348f314812b6971b26
3
+ size 116779
cache_dir/glue_stsb_train_sentence2/node_figure.json ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:71e3fc5cad7657e66a89a3ecee0777343a776ed5a18433c07004c284b50683c2
3
+ size 40366
cache_dir/glue_stsb_train_sentence2/node_list.th ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:1e9b037c3fc571116eda72dafbc56472ebba51239dd4f7fcbc271b5cb7ad238c
3
+ size 731929
cache_dir/glue_stsb_validation_sentence1/node_figure.json ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:8981d692a18cda7b99a2e716243b2f18de829113fd2c85843aa4e2c01df23021
3
+ size 23018
cache_dir/glue_stsb_validation_sentence1/node_list.th ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:db38ffd801f0a9166e88b2942bd030a5e68eae1f1ad9977c0d40af96eb00aa52
3
+ size 227053
cache_dir/glue_stsb_validation_sentence2/node_figure.json ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:a7f660fc8f3f6717e4a74aeb1489949e934ce23945424e684d936bb07f7ea14c
3
+ size 24054
cache_dir/glue_stsb_validation_sentence2/node_list.th ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:889fe5d2eeb6acfe724d8fb7fa5745db08789c03490ac5af85c8ba63879a312e
3
+ size 249945
cache_dir/glue_wnli_validation_sentence1/node_list.th ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:a7887a92b8e33bc541f85520a8a6061e269e67fc149a304eda27e0cc1eb3fcf1
3
+ size 52369
cache_dir/glue_wnli_validation_sentence2/node_list.th ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:c572bbdcc9100a87de5ffd11d02daeb92f20b5ecba15b42e024a91e7ef2c9512
3
+ size 56075
cache_dir/hate_speech18_default_train_text/fig_tok_length.png ADDED

Git LFS Details

  • SHA256: 2f796e3621124b00c94bdc833142bb1f0b3108af3f5ff28493e59b09e8a832c9
  • Pointer size: 130 Bytes
  • Size of remote file: 44.5 kB
cache_dir/hate_speech18_default_train_text/node_list.th ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:2f9efd628c345e592332bec6814760e8ec42c9f9fdea5203fd3ee53d00da4b2a
3
+ size 780029
cache_dir/hate_speech_offensive_default_train_tweet/node_list.th ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:1873ccd21b76d2f635e6ff2e945370dd206b640fed9282377fc3be38893e8b90
3
+ size 2774137
cache_dir/squad_plain_text_train_context/node_figure.json ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:843d2c9900094968bf3301adfeb2362da9088fade5db6eafc7982dcfa2f60958
3
+ size 376828
cache_dir/squad_plain_text_train_context/node_list.th ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:2e36c87b490d2b09e79fbd24dc8fb24b52333e60c9476a0a79d081d85aa93424
3
+ size 12505231
cache_dir/squad_plain_text_train_question/node_figure.json ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:765ee4f723cd42c95deb015939a7197ed2d18e344b73a12a10034d9580d0ebbc
3
+ size 86555
cache_dir/squad_plain_text_train_question/node_list.th ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:0a3ec8af3ecba7f4374769aef8ab51debe52018c99bcdfad3fd729a007ba898c
3
+ size 13804669
cache_dir/squad_plain_text_train_title/node_figure.json ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:7ee3825fb47a68afc0719467283545aa5724a1f28b872a4dcc34d986afff650a
3
+ size 45128
cache_dir/squad_plain_text_train_title/node_list.th ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:7d8954d6546bf071b93476abc3c6a588191e7d9c7efe7c1128d7deb546859d02
3
+ size 9737353
cache_dir/squad_plain_text_validation_context/node_figure.json ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:9831208533973c352bacbfb2606e6fd528ed5b8ebab283bde4117ec57e10ff3d
3
+ size 146820
cache_dir/squad_plain_text_validation_context/node_list.th ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:c1a606b7e88f0a1bbef8bb2235ef14cb27df28d3dd2f52d15d2930679f7e3f1e
3
+ size 2613513
cache_dir/squad_plain_text_validation_question/node_figure.json ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:756135d08e85affab9877fd61de96b807cadbd45c4887c29c064313e213c5fa1
3
+ size 51914
cache_dir/squad_plain_text_validation_question/node_list.th ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:07252cc5b4a44e5d3b5add905c24c43c929a2c009a217dc9262fcc580182b27f
3
+ size 1807477
cache_dir/squad_plain_text_validation_title/node_figure.json ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:d81644ec742d6a8285ac8e0fdba02759a4b582364d976cbb0c611f17c51988d5
3
+ size 11878
cache_dir/squad_plain_text_validation_title/node_list.th ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:7ea82a84a4b08e9db68c57d986554ba10cbc82eb174a6603cbb317a897c001fb
3
+ size 816063
cache_dir/squad_v2_squad_v2_train_context/node_figure.json ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:b8879eccd3e953f98f985f790a5be614a6961038dc984bdbcb22d67cb8c8ac2f
3
+ size 541497
cache_dir/squad_v2_squad_v2_train_context/node_list.th ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:109ea4f506b855fe77461be0ca4cbd65a6b98c39e3dca33cfce02792afe3724a
3
+ size 20083525
cache_dir/squad_v2_squad_v2_train_question/node_figure.json ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:35bbea47eeb55e86f204112d6e61a267285a76ae1fe822008e3037e1d6d07d7d
3
+ size 99362
cache_dir/squad_v2_squad_v2_train_question/node_list.th ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:61f14d2114169ee5eba6b0b51c0f39bee7e5f75c51afc5770a039d14a76f3843
3
+ size 21603689
cache_dir/squad_v2_squad_v2_train_title/node_figure.json ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:ce9fb6c63c4256536a8a80fa04bcc6f2bf08b99a2be28f42766eaf51df0f6ca1
3
+ size 45935
cache_dir/squad_v2_squad_v2_train_title/node_list.th ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:b1bce030d87e975f45ceb0e8beef51321724f14b8e58c27637d79adf06c2ae24
3
+ size 13959981
cache_dir/squad_v2_squad_v2_validation_context/node_figure.json ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:7a0824b9e77f79dcd0f7e30ea65378652573e19a6f29b8185ecd4829877dfccc
3
+ size 156410
cache_dir/squad_v2_squad_v2_validation_context/node_list.th ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:767ccc25a3219c5217cf326a4450b3d3216268039c9906f847e7954fc79d42f4
3
+ size 3628581
cache_dir/squad_v2_squad_v2_validation_question/node_figure.json ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:e75dda24fea60a806cd7237e1088393a39921ec8db43735b21bb6bd5f8104690
3
+ size 46101
cache_dir/squad_v2_squad_v2_validation_question/node_list.th ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:56253999490ad24ae5ee6ea374e6e719b2afad3aee4989bae046c26e97ade127
3
+ size 1935861
cache_dir/squad_v2_squad_v2_validation_title/node_figure.json ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:52a3670d5d554be48c9f9f38e6ed98460b778c8deebd69f18af7e0b4487a7272
3
+ size 10482
cache_dir/squad_v2_squad_v2_validation_title/node_list.th ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:2f438f223ce5e5088fa55986f9d857bc04247afbf3251e1dcf88a0b71f418cc5
3
+ size 778963
cache_dir/super_glue_boolq_train_passage/node_figure.json ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:499362ddcccbc575d7d129805a34daad87ed9b103f4225861a49e499cd3714dd
3
+ size 91536