Lucasgitguy-co commited on
Commit
eb8d17d
·
verified ·
1 Parent(s): 39ca201

wiki-model 0.1.0

Browse files
.gitattributes CHANGED
@@ -58,3 +58,5 @@ saved_model/**/* filter=lfs diff=lfs merge=lfs -text
58
  # Video files - compressed
59
  *.mp4 filter=lfs diff=lfs merge=lfs -text
60
  *.webm filter=lfs diff=lfs merge=lfs -text
 
 
 
58
  # Video files - compressed
59
  *.mp4 filter=lfs diff=lfs merge=lfs -text
60
  *.webm filter=lfs diff=lfs merge=lfs -text
61
+ index/bm25/vocab.index.json filter=lfs diff=lfs merge=lfs -text
62
+ index/docs.sqlite filter=lfs diff=lfs merge=lfs -text
README.md ADDED
@@ -0,0 +1,44 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ ---
2
+ license: cc-by-sa-4.0
3
+ language: [en]
4
+ pretty_name: wiki-model facts
5
+ tags: [wikipedia, facts, question-answering, retrieval, bm25]
6
+ size_categories: [1M<n<10M]
7
+ source_datasets: [wikimedia/structured-wikipedia]
8
+ configs:
9
+ - config_name: tasks
10
+ data_files:
11
+ - {split: train, path: tasks/train.jsonl.gz}
12
+ - {split: validation, path: tasks/val.jsonl.gz}
13
+ - {split: test, path: tasks/test.jsonl.gz}
14
+ - config_name: articles
15
+ data_files: articles/wiki.jsonl.gz
16
+ ---
17
+
18
+ # wiki-model facts
19
+
20
+ Training data and the sourced fact index for [wiki-model](https://huggingface.co/Lucasgitguy-co/wiki-model) and
21
+ [wiki-model-tiny](https://huggingface.co/Lucasgitguy-co/wiki-model-tiny), built from
22
+ [wikimedia/structured-wikipedia](https://huggingface.co/datasets/wikimedia/structured-wikipedia) (English, 2024).
23
+
24
+ | Path | What |
25
+ |---|---|
26
+ | `articles/wiki.jsonl.gz` | 947,727 articles: `title, url, desc, sents` (up to 3 lead sentences), `fields` (selected infobox facts), `src` |
27
+ | `tasks/{train,val,test}.jsonl.gz` | `task` (fact / q / fix), `title`, `prompt`, `target`, `ctx` (context text or null). Split by article title. |
28
+ | `index/` | bm25s index (`bm25/`) + `docs.sqlite` (article text, zlib) used by `--grounded` |
29
+
30
+ Articles: 232k random articles from three shards, plus 716k long articles (at least 15k characters, no digits in the title) from the other 83 shards.
31
+
32
+ <details><summary>How tasks are built</summary>
33
+
34
+ - **fact**: `<fact> Title </fact>` → the first one or two lead sentences.
35
+ - **q**: `<q> What is Title? </q>` → the short description; `<q> When was Title founded? </q>` etc. → an infobox value (about 60 fields).
36
+ - **fix**: an abstract sentence with a year shifted by 1–30 years → the original; or an unchanged sentence → itself.
37
+ - **ctx**: 35% of examples carry `<ctx>` text: the article's own index text (same format as the index), half the time together with a random distractor article. At run time `--grounded` fills it from BM25 search instead.
38
+ - The models were also trained on [Bulutthecat/FactfulDataset](https://huggingface.co/datasets/Bulutthecat/FactfulDataset) fix examples. They are **not included here**, because that dataset has no licence on its card.
39
+ </details>
40
+
41
+ ## Licence and attribution
42
+
43
+ CC BY-SA 4.0. Text and data come from Wikipedia, written by Wikipedia contributors, via the Wikimedia Enterprise structured contents
44
+ snapshot. Each article row keeps its source URL.
articles/wiki.jsonl.gz ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:67491ecbe8096f5b530ad0ca1f87f2085192d30b3cfe54ece8cd6dc369cf384d
3
+ size 187400124
index/bm25/data.csc.index.npy ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:cf2a70387babe2a8b332901529ac02cfc42550f77e73b47f3c9b8d96cc5a165a
3
+ size 120200560
index/bm25/indices.csc.index.npy ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:172ec84ef5042cb40fcadde44c0b58ab35f86fb5bc4f382b570311e83e0bc099
3
+ size 120200560
index/bm25/indptr.csc.index.npy ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:4fb5787fd45317b694a2089db4d4ce1186d3a877c4c6a0b836bc0211d677ab5e
3
+ size 5920176
index/bm25/params.index.json ADDED
@@ -0,0 +1,12 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "k1": 1.5,
3
+ "b": 0.75,
4
+ "delta": 0.5,
5
+ "method": "lucene",
6
+ "idf_method": "lucene",
7
+ "dtype": "float32",
8
+ "int_dtype": "int32",
9
+ "num_docs": 947727,
10
+ "version": "0.3.16",
11
+ "backend": "numpy"
12
+ }
index/bm25/vocab.index.json ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:b81b79eca039521955dd3a0124770da1585969403fd570c1c5cebfe2a3c1d73b
3
+ size 14535888
index/docs.sqlite ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:e93910a801ad4c88b7e67d007c330b99e94590160cce276ba7176747d3aa4ed1
3
+ size 338231296
tasks/test.jsonl.gz ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:03bf4770c6c3f3faa0bf3a853718526af8f7e644672f7cfc6491b843fe3d1dd4
3
+ size 5180735
tasks/train.jsonl.gz ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:789d4ac4012260b54c5ca1c88ea114c9ad7afde825ebd07c4a445a7914f64ffe
3
+ size 247374235
tasks/val.jsonl.gz ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:3ec4fcbd7866983d37b2ad1273288d80ecc044cd58ab7c62a976f4f5e2d28926
3
+ size 2559897