Hugging Face's logo

Spaces:
huggingface
/
data-measurements-tool
Running on T4

App Files Community
4
Ezi Ozoani commited on
Commit
3d4f393
•
1 Parent(s): 06e221a

All changes minus temp.jsonl from last 3 commits

Browse files
This view is limited to 50 files because it contains too many changes.   See raw diff
Files changed (50) hide show
  1. .DS_Store +0 -0
  2. .gitattributes +2 -1
  3. README.md +1 -1
  4. cache_dir/.DS_Store +1 -1
  5. cache_dir/HuggingFaceM4/.DS_Store +1 -1
  6. cache_dir/HuggingFaceM4/{OBELICS_opt_out_docs_removed_2023_07_12_train_general_metadata/general_stats_dict.json → OBELICS_default_train_text/.DS_Store} +2 -2
  7. cache_dir/HuggingFaceM4/{OBELICS_default_train_texts → OBELICS_default_train_text}/associations/identity_terms.json +0 -0
  8. cache_dir/HuggingFaceM4/{OBELICS_default_train_texts → OBELICS_default_train_text/base_dset}/.DS_Store +1 -1
  9. cache_dir/HuggingFaceM4/{OBELICS_default_train_texts → OBELICS_default_train_text}/base_dset/data-00000-of-00002.arrow +0 -0
  10. cache_dir/HuggingFaceM4/{OBELICS_default_train_texts → OBELICS_default_train_text}/base_dset/data-00001-of-00002.arrow +0 -0
  11. cache_dir/HuggingFaceM4/{OBELICS_default_train_texts → OBELICS_default_train_text}/base_dset/dataset_info.json +0 -0
  12. cache_dir/HuggingFaceM4/{OBELICS_default_train_texts → OBELICS_default_train_text}/base_dset/state.json +0 -0
  13. cache_dir/HuggingFaceM4/{OBELICS_default_train_texts → OBELICS_default_train_text}/general_stats_dict.json +0 -0
  14. cache_dir/HuggingFaceM4/{OBELICS_default_train_texts → OBELICS_default_train_text}/lengths/length_measurements.json +0 -0
  15. cache_dir/HuggingFaceM4/{OBELICS_default_train_texts → OBELICS_default_train_text}/lengths/lengths_fig.png +0 -0
  16. cache_dir/HuggingFaceM4/{OBELICS_default_train_texts → OBELICS_default_train_text}/lengths/lengths_table.json +0 -0
  17. cache_dir/HuggingFaceM4/{OBELICS_default_train_texts → OBELICS_default_train_text}/sorted_top_vocab.json +0 -0
  18. cache_dir/HuggingFaceM4/{OBELICS_default_train_texts → OBELICS_default_train_text}/text_dset/data-00000-of-00001.arrow +0 -0
  19. cache_dir/HuggingFaceM4/{OBELICS_default_train_texts → OBELICS_default_train_text}/text_dset/dataset_info.json +0 -0
  20. cache_dir/HuggingFaceM4/{OBELICS_default_train_texts → OBELICS_default_train_text}/text_dset/state.json +0 -0
  21. cache_dir/HuggingFaceM4/{OBELICS_default_train_texts → OBELICS_default_train_text}/text_duplicates/text_duplicates.html +0 -0
  22. cache_dir/HuggingFaceM4/{OBELICS_default_train_texts → OBELICS_default_train_text}/text_duplicates/text_duplicates.json +0 -0
  23. cache_dir/HuggingFaceM4/{OBELICS_default_train_texts → OBELICS_default_train_text}/tokenized_df.json +0 -0
  24. cache_dir/HuggingFaceM4/{OBELICS_default_train_texts → OBELICS_default_train_text}/vocab_counts.json +0 -0
  25. cache_dir/HuggingFaceM4/OBELICS_opt_out_docs_removed_2023_07_12_train_general_metadata/base_dset/data-00000-of-00001.arrow +0 -3
  26. cache_dir/HuggingFaceM4/OBELICS_opt_out_docs_removed_2023_07_12_train_general_metadata/base_dset/dataset_info.json +0 -3
  27. cache_dir/HuggingFaceM4/OBELICS_opt_out_docs_removed_2023_07_12_train_general_metadata/base_dset/state.json +0 -3
  28. cache_dir/HuggingFaceM4/OBELICS_opt_out_docs_removed_2023_07_12_train_general_metadata/dset_peek.json +0 -3
  29. cache_dir/HuggingFaceM4/OBELICS_opt_out_docs_removed_2023_07_12_train_general_metadata/lengths/length_measurements.json +0 -3
  30. cache_dir/HuggingFaceM4/OBELICS_opt_out_docs_removed_2023_07_12_train_general_metadata/lengths/lengths_fig.png +0 -3
  31. cache_dir/HuggingFaceM4/OBELICS_opt_out_docs_removed_2023_07_12_train_general_metadata/lengths/lengths_table.json +0 -3
  32. cache_dir/HuggingFaceM4/OBELICS_opt_out_docs_removed_2023_07_12_train_general_metadata/sorted_top_vocab.json +0 -3
  33. cache_dir/HuggingFaceM4/OBELICS_opt_out_docs_removed_2023_07_12_train_general_metadata/text_dset/cache-f6aa4a70e38b4a04.arrow +0 -3
  34. cache_dir/HuggingFaceM4/OBELICS_opt_out_docs_removed_2023_07_12_train_general_metadata/text_dset/data-00000-of-00001.arrow +0 -3
  35. cache_dir/HuggingFaceM4/OBELICS_opt_out_docs_removed_2023_07_12_train_general_metadata/text_dset/dataset_info.json +0 -3
  36. cache_dir/HuggingFaceM4/OBELICS_opt_out_docs_removed_2023_07_12_train_general_metadata/text_dset/state.json +0 -3
  37. cache_dir/HuggingFaceM4/OBELICS_opt_out_docs_removed_2023_07_12_train_general_metadata/text_duplicates/text_duplicates.html +0 -3
  38. cache_dir/HuggingFaceM4/OBELICS_opt_out_docs_removed_2023_07_12_train_general_metadata/text_duplicates/text_duplicates.json +0 -3
  39. cache_dir/HuggingFaceM4/OBELICS_opt_out_docs_removed_2023_07_12_train_general_metadata/tokenized_df.json +0 -3
  40. cache_dir/HuggingFaceM4/OBELICS_opt_out_docs_removed_2023_07_12_train_general_metadata/vocab_counts.json +0 -3
  41. cache_dir/HuggingFaceM4/OBELICS_opt_out_docs_removed_2023_07_12_train_general_metadata/zipf/zipf_basic_stats.json +0 -3
  42. cache_dir/HuggingFaceM4/OBELICS_opt_out_docs_removed_2023_07_12_train_general_metadata/zipf/zipf_fig.html +0 -3
  43. cache_dir/HuggingFaceM4/OBELICS_opt_out_docs_removed_2023_07_12_train_general_metadata/zipf/zipf_fig.json +0 -3
  44. cache_dir/HuggingFaceM4/OBELICS_opt_out_docs_removed_2023_07_12_train_images/base_dset/data-00000-of-00001.arrow +0 -3
  45. cache_dir/HuggingFaceM4/OBELICS_opt_out_docs_removed_2023_07_12_train_images/base_dset/dataset_info.json +0 -3
  46. cache_dir/HuggingFaceM4/OBELICS_opt_out_docs_removed_2023_07_12_train_images/base_dset/state.json +0 -3
  47. cache_dir/HuggingFaceM4/OBELICS_opt_out_docs_removed_2023_07_12_train_images/dset_peek.json +0 -3
  48. cache_dir/HuggingFaceM4/OBELICS_opt_out_docs_removed_2023_07_12_train_images/general_stats_dict.json +0 -3
  49. cache_dir/HuggingFaceM4/OBELICS_opt_out_docs_removed_2023_07_12_train_images/lengths/length_measurements.json +0 -3
  50. cache_dir/HuggingFaceM4/OBELICS_opt_out_docs_removed_2023_07_12_train_images/lengths/lengths_fig.png +0 -3
.DS_Store CHANGED
Binary files a/.DS_Store and b/.DS_Store differ
.gitattributes CHANGED
@@ -33,4 +33,5 @@ saved_model/**/* filter=lfs diff=lfs merge=lfs -text
33
  *.mov filter=lfs diff=lfs merge=lfs -text
34
  *.gif filter=lfs diff=lfs merge=lfs -text
35
  cache_dir/**/* filter=lfs diff=lfs merge=lfs -text
36
- cache_dir/**/**/* filter=lfs diff=lfs merge=lfs -text
 
33
  *.mov filter=lfs diff=lfs merge=lfs -text
34
  *.gif filter=lfs diff=lfs merge=lfs -text
35
  cache_dir/**/* filter=lfs diff=lfs merge=lfs -text
36
+ cache_dir/**/**/* filter=lfs diff=lfs merge=lfs -text
37
+ *.jsonl filter=lfs diff=lfs merge=lfs -text
README.md CHANGED
@@ -7,7 +7,7 @@ sdk: streamlit
7
  sdk_version: 1.0.0
8
  app_file: app.py
9
  pinned: false
10
- python_version: 3.9
11
  ---
12
 
13
  # Data Measurements Tool
7
  sdk_version: 1.0.0
8
  app_file: app.py
9
  pinned: false
10
+ python_version: 3.9.6
11
  ---
12
 
13
  # Data Measurements Tool
cache_dir/.DS_Store CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:5f54406681cc569a9d56248a3be247afb7952fa5272ab20bb039e584f80d26c6
3
  size 14340
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:19c39a390bcad50240ee311709c795ccc781d245bbc0177fec7b798e4aa0f86a
3
  size 14340
cache_dir/HuggingFaceM4/.DS_Store CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:3eda935b3db4b738eb991f2ad964f1f4f075f9161bf3df13fb43c6a7a0fda6e1
3
  size 6148
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:3e21720a64cd789a9d8e7c6369628d2ba25057fde8d2394a45e8f798fb71f14d
3
  size 6148
cache_dir/HuggingFaceM4/{OBELICS_opt_out_docs_removed_2023_07_12_train_general_metadata/general_stats_dict.json → OBELICS_default_train_text/.DS_Store} RENAMED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:c7a45931e78871dcb39b814081ef2c49939709c3feebcddaf7dc9e221a557b24
3
- size 95
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:7057786add5ec9a6a3fac1e38d1fd334a8e3006ed45e10564a7957bed82a627f
3
+ size 8196
cache_dir/HuggingFaceM4/{OBELICS_default_train_texts → OBELICS_default_train_text}/associations/identity_terms.json RENAMED
File without changes
cache_dir/HuggingFaceM4/{OBELICS_default_train_texts → OBELICS_default_train_text/base_dset}/.DS_Store RENAMED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:2f40ba1ae0df6f456333e7b4708a4b420f29a8910a9f93dea56797202bb4b8d2
3
  size 6148
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:d65165279105ca6773180500688df4bdc69a2c7b771752f0a46ef120b7fd8ec3
3
  size 6148
cache_dir/HuggingFaceM4/{OBELICS_default_train_texts → OBELICS_default_train_text}/base_dset/data-00000-of-00002.arrow RENAMED
File without changes
cache_dir/HuggingFaceM4/{OBELICS_default_train_texts → OBELICS_default_train_text}/base_dset/data-00001-of-00002.arrow RENAMED
File without changes
cache_dir/HuggingFaceM4/{OBELICS_default_train_texts → OBELICS_default_train_text}/base_dset/dataset_info.json RENAMED
File without changes
cache_dir/HuggingFaceM4/{OBELICS_default_train_texts → OBELICS_default_train_text}/base_dset/state.json RENAMED
File without changes
cache_dir/HuggingFaceM4/{OBELICS_default_train_texts → OBELICS_default_train_text}/general_stats_dict.json RENAMED
File without changes
cache_dir/HuggingFaceM4/{OBELICS_default_train_texts → OBELICS_default_train_text}/lengths/length_measurements.json RENAMED
File without changes
cache_dir/HuggingFaceM4/{OBELICS_default_train_texts → OBELICS_default_train_text}/lengths/lengths_fig.png RENAMED
File without changes
cache_dir/HuggingFaceM4/{OBELICS_default_train_texts → OBELICS_default_train_text}/lengths/lengths_table.json RENAMED
File without changes
cache_dir/HuggingFaceM4/{OBELICS_default_train_texts → OBELICS_default_train_text}/sorted_top_vocab.json RENAMED
File without changes
cache_dir/HuggingFaceM4/{OBELICS_default_train_texts → OBELICS_default_train_text}/text_dset/data-00000-of-00001.arrow RENAMED
File without changes
cache_dir/HuggingFaceM4/{OBELICS_default_train_texts → OBELICS_default_train_text}/text_dset/dataset_info.json RENAMED
File without changes
cache_dir/HuggingFaceM4/{OBELICS_default_train_texts → OBELICS_default_train_text}/text_dset/state.json RENAMED
File without changes
cache_dir/HuggingFaceM4/{OBELICS_default_train_texts → OBELICS_default_train_text}/text_duplicates/text_duplicates.html RENAMED
File without changes
cache_dir/HuggingFaceM4/{OBELICS_default_train_texts → OBELICS_default_train_text}/text_duplicates/text_duplicates.json RENAMED
File without changes
cache_dir/HuggingFaceM4/{OBELICS_default_train_texts → OBELICS_default_train_text}/tokenized_df.json RENAMED
File without changes
cache_dir/HuggingFaceM4/{OBELICS_default_train_texts → OBELICS_default_train_text}/vocab_counts.json RENAMED
File without changes
cache_dir/HuggingFaceM4/OBELICS_opt_out_docs_removed_2023_07_12_train_general_metadata/base_dset/data-00000-of-00001.arrow DELETED
@@ -1,3 +0,0 @@
1
- version https://git-lfs.github.com/spec/v1
2
- oid sha256:2f23eb11877f1bf934ba12d0ce910ccb71b9ce3865798dc6651d12425244b529
3
- size 489144
 
 
 
cache_dir/HuggingFaceM4/OBELICS_opt_out_docs_removed_2023_07_12_train_general_metadata/base_dset/dataset_info.json DELETED
@@ -1,3 +0,0 @@
1
- version https://git-lfs.github.com/spec/v1
2
- oid sha256:bbbef344f84ed9a7f11d0778f52ff0e2decf2ef7aacd1fbf5f059d069d2c6ba6
3
- size 946
 
 
 
cache_dir/HuggingFaceM4/OBELICS_opt_out_docs_removed_2023_07_12_train_general_metadata/base_dset/state.json DELETED
@@ -1,3 +0,0 @@
1
- version https://git-lfs.github.com/spec/v1
2
- oid sha256:ab2c18f9e7e23bc6b501df499e16c721198a5452b993cbc3921612254438a7c3
3
- size 250
 
 
 
cache_dir/HuggingFaceM4/OBELICS_opt_out_docs_removed_2023_07_12_train_general_metadata/dset_peek.json DELETED
@@ -1,3 +0,0 @@
1
- version https://git-lfs.github.com/spec/v1
2
- oid sha256:c8be5416e6a1bf0c3d02709b3d5c13b7b9af543ea902301e5b268b4d16b026c0
3
- size 502887
 
 
 
cache_dir/HuggingFaceM4/OBELICS_opt_out_docs_removed_2023_07_12_train_general_metadata/lengths/length_measurements.json DELETED
@@ -1,3 +0,0 @@
1
- version https://git-lfs.github.com/spec/v1
2
- oid sha256:500e2c7bd48b76556389af8630b5093b147c32affc0528305696531841df7582
3
- size 113
 
 
 
cache_dir/HuggingFaceM4/OBELICS_opt_out_docs_removed_2023_07_12_train_general_metadata/lengths/lengths_fig.png DELETED

Git LFS Details

  • SHA256: c4136b1d8653ecb9b425b162624c8adc8ca2c53ae44af90a34b65b0aaa946fb1
  • Pointer size: 130 Bytes
  • Size of remote file: 69.9 kB
cache_dir/HuggingFaceM4/OBELICS_opt_out_docs_removed_2023_07_12_train_general_metadata/lengths/lengths_table.json DELETED
@@ -1,3 +0,0 @@
1
- version https://git-lfs.github.com/spec/v1
2
- oid sha256:3d585449245038b71c834729d34b61b023b9866eec20f39fe33430a97a84be38
3
- size 33277
 
 
 
cache_dir/HuggingFaceM4/OBELICS_opt_out_docs_removed_2023_07_12_train_general_metadata/sorted_top_vocab.json DELETED
@@ -1,3 +0,0 @@
1
- version https://git-lfs.github.com/spec/v1
2
- oid sha256:6a64574c2dd6522190927d73577c84e293e11f179f37482e2305a159d27e7e80
3
- size 7785
 
 
 
cache_dir/HuggingFaceM4/OBELICS_opt_out_docs_removed_2023_07_12_train_general_metadata/text_dset/cache-f6aa4a70e38b4a04.arrow DELETED
@@ -1,3 +0,0 @@
1
- version https://git-lfs.github.com/spec/v1
2
- oid sha256:07e5fe498e11eeb301bd8430d2f2c408c696d36bbc21b1d18f31c4cf957a8d47
3
- size 67576
 
 
 
cache_dir/HuggingFaceM4/OBELICS_opt_out_docs_removed_2023_07_12_train_general_metadata/text_dset/data-00000-of-00001.arrow DELETED
@@ -1,3 +0,0 @@
1
- version https://git-lfs.github.com/spec/v1
2
- oid sha256:d04dc98da752f00f252635eed55717c8a12dee20edb3c8347bd34e2228b16bb3
3
- size 29512
 
 
 
cache_dir/HuggingFaceM4/OBELICS_opt_out_docs_removed_2023_07_12_train_general_metadata/text_dset/dataset_info.json DELETED
@@ -1,3 +0,0 @@
1
- version https://git-lfs.github.com/spec/v1
2
- oid sha256:362e98087a1b5bad2dc4e4d4b04dde74ed38ab64be0c12722e1f328518055d3b
3
- size 604
 
 
 
cache_dir/HuggingFaceM4/OBELICS_opt_out_docs_removed_2023_07_12_train_general_metadata/text_dset/state.json DELETED
@@ -1,3 +0,0 @@
1
- version https://git-lfs.github.com/spec/v1
2
- oid sha256:5589b05d616fd17ead2fd469ad921b88ee2c464cb5bd21afca6a4876403c1ba7
3
- size 250
 
 
 
cache_dir/HuggingFaceM4/OBELICS_opt_out_docs_removed_2023_07_12_train_general_metadata/text_duplicates/text_duplicates.html DELETED
@@ -1,3 +0,0 @@
1
- version https://git-lfs.github.com/spec/v1
2
- oid sha256:b1690a778ad39a5363a94de88f59dda4fcf5ffbf4c3cc5abfbddbbc3be927fc0
3
- size 116
 
 
 
cache_dir/HuggingFaceM4/OBELICS_opt_out_docs_removed_2023_07_12_train_general_metadata/text_duplicates/text_duplicates.json DELETED
@@ -1,3 +0,0 @@
1
- version https://git-lfs.github.com/spec/v1
2
- oid sha256:c40970be1f0326d8ef1b689eda4ac2695d0c536d95fc69cfcd569cb361a77ad6
3
- size 50
 
 
 
cache_dir/HuggingFaceM4/OBELICS_opt_out_docs_removed_2023_07_12_train_general_metadata/tokenized_df.json DELETED
@@ -1,3 +0,0 @@
1
- version https://git-lfs.github.com/spec/v1
2
- oid sha256:d68e3413893f2173dc0abf1c0f818d51b14c5c1618622b5ba31d903423df3a21
3
- size 71186
 
 
 
cache_dir/HuggingFaceM4/OBELICS_opt_out_docs_removed_2023_07_12_train_general_metadata/vocab_counts.json DELETED
@@ -1,3 +0,0 @@
1
- version https://git-lfs.github.com/spec/v1
2
- oid sha256:fc2dbb4848f61dcc440ce6a940617092309b10c5dd589517772509102dc4b88f
3
- size 121825
 
 
 
cache_dir/HuggingFaceM4/OBELICS_opt_out_docs_removed_2023_07_12_train_general_metadata/zipf/zipf_basic_stats.json DELETED
@@ -1,3 +0,0 @@
1
- version https://git-lfs.github.com/spec/v1
2
- oid sha256:b56322c6815bf2be43a1d03dc8dd20c76c8ddf96e68a719e95c92d07917b06e3
3
- size 329
 
 
 
cache_dir/HuggingFaceM4/OBELICS_opt_out_docs_removed_2023_07_12_train_general_metadata/zipf/zipf_fig.html DELETED
@@ -1,3 +0,0 @@
1
- version https://git-lfs.github.com/spec/v1
2
- oid sha256:0f188fe76c2135208a74b9c5bfb7e107d3473f513205489ade28cb0e727adc42
3
- size 3606637
 
 
 
cache_dir/HuggingFaceM4/OBELICS_opt_out_docs_removed_2023_07_12_train_general_metadata/zipf/zipf_fig.json DELETED
@@ -1,3 +0,0 @@
1
- version https://git-lfs.github.com/spec/v1
2
- oid sha256:1a49d21ab8319d8a2d9177770b511b2b31b735c70a28f89afeb140b5c6f938ff
3
- size 20320
 
 
 
cache_dir/HuggingFaceM4/OBELICS_opt_out_docs_removed_2023_07_12_train_images/base_dset/data-00000-of-00001.arrow DELETED
@@ -1,3 +0,0 @@
1
- version https://git-lfs.github.com/spec/v1
2
- oid sha256:2f23eb11877f1bf934ba12d0ce910ccb71b9ce3865798dc6651d12425244b529
3
- size 489144
 
 
 
cache_dir/HuggingFaceM4/OBELICS_opt_out_docs_removed_2023_07_12_train_images/base_dset/dataset_info.json DELETED
@@ -1,3 +0,0 @@
1
- version https://git-lfs.github.com/spec/v1
2
- oid sha256:bbbef344f84ed9a7f11d0778f52ff0e2decf2ef7aacd1fbf5f059d069d2c6ba6
3
- size 946
 
 
 
cache_dir/HuggingFaceM4/OBELICS_opt_out_docs_removed_2023_07_12_train_images/base_dset/state.json DELETED
@@ -1,3 +0,0 @@
1
- version https://git-lfs.github.com/spec/v1
2
- oid sha256:ab2c18f9e7e23bc6b501df499e16c721198a5452b993cbc3921612254438a7c3
3
- size 250
 
 
 
cache_dir/HuggingFaceM4/OBELICS_opt_out_docs_removed_2023_07_12_train_images/dset_peek.json DELETED
@@ -1,3 +0,0 @@
1
- version https://git-lfs.github.com/spec/v1
2
- oid sha256:c8be5416e6a1bf0c3d02709b3d5c13b7b9af543ea902301e5b268b4d16b026c0
3
- size 502887
 
 
 
cache_dir/HuggingFaceM4/OBELICS_opt_out_docs_removed_2023_07_12_train_images/general_stats_dict.json DELETED
@@ -1,3 +0,0 @@
1
- version https://git-lfs.github.com/spec/v1
2
- oid sha256:51c5e8fe33ca5024e519e7d550f46e3917c9ba8dc63f43ecc107c3c31430cb00
3
- size 93
 
 
 
cache_dir/HuggingFaceM4/OBELICS_opt_out_docs_removed_2023_07_12_train_images/lengths/length_measurements.json DELETED
@@ -1,3 +0,0 @@
1
- version https://git-lfs.github.com/spec/v1
2
- oid sha256:0889a390641bd052cbf295478822db4509a1731d20f7fda753198c64f9970c88
3
- size 125
 
 
 
cache_dir/HuggingFaceM4/OBELICS_opt_out_docs_removed_2023_07_12_train_images/lengths/lengths_fig.png DELETED

Git LFS Details

  • SHA256: 87e075d701bdeacb58f17fab26d2e4a98e861be2a5e67282b9fa0fabb589181e
  • Pointer size: 130 Bytes
  • Size of remote file: 66.3 kB