tamnd commited on
Commit
4577713
·
verified ·
1 Parent(s): 9508bad

Publish shard CC-MAIN-2026-08/09946

Browse files
Files changed (3) hide show
  1. README.md +10 -10
  2. data/CC-MAIN-2026-08/09946.parquet +3 -0
  3. stats.csv +2 -1
README.md CHANGED
@@ -32,7 +32,7 @@ configs:
32
 
33
  Open Index is a large-scale web text dataset built from [Common Crawl](https://commoncrawl.org). Every page goes through a pipeline that extracts the main content from raw HTML, converts it to clean Markdown using [trafilatura](https://github.com/adbar/trafilatura), and packages the result into Parquet files with full WARC metadata preserved.
34
 
35
- The dataset currently includes crawl **CC-MAIN-2026-08** with **188,548,954 documents across 9999 shards**. We plan to add more snapshots over time.
36
 
37
  Open Index is released under the **Open Data Commons Attribution License (ODC-By) v1.0**, the same license used by Common Crawl.
38
 
@@ -164,28 +164,28 @@ Empty conversions (pages where trafilatura could not extract meaningful content)
164
 
165
  ### Compression Ratios
166
 
167
- Numbers below are actual measurements summed across all 9999 files of CC-MAIN-2026-08 (188,548,954 pages total), projected to the full crawl of 100,000 WARC files.
168
 
169
- | Stage | 9999 files (measured) | 100,000 files (projected) | Reduction |
170
  |---|---|---|---|
171
  | Raw WARC (.warc.gz, downloaded) | ~7.9 TB | ~83 TB | — |
172
- | HTML extracted (uncompressed) | 22.6 TB | ~295 TB | — |
173
  | Packed markdown WARC (.md.warc.gz) | ~396.8 GB | ~3.7 TB | **-98.3%** vs HTML |
174
  | Final Parquet (Zstd level 19) | 267.7 GB | ~2.9 TB | **-32.5%** vs packed WARC |
175
 
176
- The big win is the HTML → Markdown step: trafilatura strips all tags, scripts, styles, navigation, and ads, keeping only the main content. This cuts 22.6 TB of uncompressed HTML down to 844.2 GB of markdown — a **98.3% reduction** — before any file-level compression is applied. Parquet with Zstd level 19 then compresses the markdown a further 68.3%.
177
 
178
- End to end: ~7.9 TB of raw gzipped WARCs becomes **267.7 GB of Parquet** — a **96.7% total reduction** — containing 188,548,954 clean markdown documents.
179
 
180
  ### Processing Times
181
 
182
- Pipeline timings across 9999 shards of CC-MAIN-2026-08:
183
 
184
  ```
185
- Download (raw WARC) ████████████████████████ total 614h 51m 49s avg 3m 41s
186
  Convert (HTML → MD) █░░░░░░░░░░░░░░░░░░░░░░░ total 17h 35m 34s avg 6s
187
- Export (Parquet) █████░░░░░░░░░░░░░░░░░░░ total 134h 51m 8s avg 48s
188
- Publish (HuggingFace) ██░░░░░░░░░░░░░░░░░░░░░░ total 66h 54m 35s avg 24s
189
  ```
190
 
191
  ### Dataset Charts
 
32
 
33
  Open Index is a large-scale web text dataset built from [Common Crawl](https://commoncrawl.org). Every page goes through a pipeline that extracts the main content from raw HTML, converts it to clean Markdown using [trafilatura](https://github.com/adbar/trafilatura), and packages the result into Parquet files with full WARC metadata preserved.
34
 
35
+ The dataset currently includes crawl **CC-MAIN-2026-08** with **188,568,328 documents across 10000 shards**. We plan to add more snapshots over time.
36
 
37
  Open Index is released under the **Open Data Commons Attribution License (ODC-By) v1.0**, the same license used by Common Crawl.
38
 
 
164
 
165
  ### Compression Ratios
166
 
167
+ Numbers below are actual measurements summed across all 10000 files of CC-MAIN-2026-08 (188,568,328 pages total), projected to the full crawl of 100,000 WARC files.
168
 
169
+ | Stage | 10000 files (measured) | 100,000 files (projected) | Reduction |
170
  |---|---|---|---|
171
  | Raw WARC (.warc.gz, downloaded) | ~7.9 TB | ~83 TB | — |
172
+ | HTML extracted (uncompressed) | 22.7 TB | ~295 TB | — |
173
  | Packed markdown WARC (.md.warc.gz) | ~396.8 GB | ~3.7 TB | **-98.3%** vs HTML |
174
  | Final Parquet (Zstd level 19) | 267.7 GB | ~2.9 TB | **-32.5%** vs packed WARC |
175
 
176
+ The big win is the HTML → Markdown step: trafilatura strips all tags, scripts, styles, navigation, and ads, keeping only the main content. This cuts 22.7 TB of uncompressed HTML down to 844.3 GB of markdown — a **98.3% reduction** — before any file-level compression is applied. Parquet with Zstd level 19 then compresses the markdown a further 68.3%.
177
 
178
+ End to end: ~7.9 TB of raw gzipped WARCs becomes **267.7 GB of Parquet** — a **96.7% total reduction** — containing 188,568,328 clean markdown documents.
179
 
180
  ### Processing Times
181
 
182
+ Pipeline timings across 10000 shards of CC-MAIN-2026-08:
183
 
184
  ```
185
+ Download (raw WARC) ████████████████████████ total 614h 53m 24s avg 3m 41s
186
  Convert (HTML → MD) █░░░░░░░░░░░░░░░░░░░░░░░ total 17h 35m 34s avg 6s
187
+ Export (Parquet) █████░░░░░░░░░░░░░░░░░░░ total 134h 51m 24s avg 48s
188
+ Publish (HuggingFace) ██░░░░░░░░░░░░░░░░░░░░░░ total 66h 54m 52s avg 24s
189
  ```
190
 
191
  ### Dataset Charts
data/CC-MAIN-2026-08/09946.parquet ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:d4e08c64d7dcc7e2f8f3b4a1796c0b9fed5ce2958b3f2da4b606aea5e4196fb2
3
+ size 30117894
stats.csv CHANGED
@@ -9946,6 +9946,7 @@ CC-MAIN-2026-08,9942,19464,2560891383,92911643,29573857,2026-03-20T03:07:47Z,147
9946
  CC-MAIN-2026-08,9943,19552,2571276485,93859920,29337690,2026-03-20T01:44:17Z,177,0,55,20
9947
  CC-MAIN-2026-08,9944,19398,2522282040,95238737,30397717,2026-03-20T01:48:17Z,149,0,62,14
9948
  CC-MAIN-2026-08,9945,19693,2576715824,95543177,29859738,2026-03-20T01:52:32Z,131,0,53,21
 
9949
  CC-MAIN-2026-08,9947,19258,2527288246,97097122,30137519,2026-03-20T01:56:02Z,244,0,28,17
9950
  CC-MAIN-2026-08,9948,19545,2574101909,95017301,29874399,2026-03-20T02:02:17Z,236,0,36,59
9951
  CC-MAIN-2026-08,9949,19429,2531040398,93121233,29400334,2026-03-20T03:34:17Z,173,0,22,21
@@ -9982,7 +9983,7 @@ CC-MAIN-2026-08,9979,19375,2541892076,91319099,29349175,2026-03-20T01:50:17Z,217
9982
  CC-MAIN-2026-08,9980,19301,2492442624,91241333,29115010,2026-03-20T03:18:32Z,121,0,18,28
9983
  CC-MAIN-2026-08,9981,19469,2582219569,94275829,29359817,2026-03-20T01:56:02Z,245,0,29,17
9984
  CC-MAIN-2026-08,9982,19365,2525200797,91288349,28770610,2026-03-20T02:02:17Z,235,0,35,59
9985
- CC-MAIN-2026-08,9983,19455,2564709111,95168529,29778127,2026-03-20T03:36:17Z,150,0,23,0
9986
  CC-MAIN-2026-08,9984,19193,2507611748,90452328,29027172,2026-03-20T02:08:02Z,239,0,34,16
9987
  CC-MAIN-2026-08,9985,19367,2531391271,94995010,30776832,2026-03-20T02:10:32Z,166,0,38,14
9988
  CC-MAIN-2026-08,9986,19589,2541080212,94152803,29609822,2026-03-20T03:23:17Z,174,0,25,50
 
9946
  CC-MAIN-2026-08,9943,19552,2571276485,93859920,29337690,2026-03-20T01:44:17Z,177,0,55,20
9947
  CC-MAIN-2026-08,9944,19398,2522282040,95238737,30397717,2026-03-20T01:48:17Z,149,0,62,14
9948
  CC-MAIN-2026-08,9945,19693,2576715824,95543177,29859738,2026-03-20T01:52:32Z,131,0,53,21
9949
+ CC-MAIN-2026-08,9946,19374,2550983301,93264377,30117894,2026-03-20T03:38:17Z,95,0,16,0
9950
  CC-MAIN-2026-08,9947,19258,2527288246,97097122,30137519,2026-03-20T01:56:02Z,244,0,28,17
9951
  CC-MAIN-2026-08,9948,19545,2574101909,95017301,29874399,2026-03-20T02:02:17Z,236,0,36,59
9952
  CC-MAIN-2026-08,9949,19429,2531040398,93121233,29400334,2026-03-20T03:34:17Z,173,0,22,21
 
9983
  CC-MAIN-2026-08,9980,19301,2492442624,91241333,29115010,2026-03-20T03:18:32Z,121,0,18,28
9984
  CC-MAIN-2026-08,9981,19469,2582219569,94275829,29359817,2026-03-20T01:56:02Z,245,0,29,17
9985
  CC-MAIN-2026-08,9982,19365,2525200797,91288349,28770610,2026-03-20T02:02:17Z,235,0,35,59
9986
+ CC-MAIN-2026-08,9983,19455,2564709111,95168529,29778127,2026-03-20T03:36:17Z,150,0,23,17
9987
  CC-MAIN-2026-08,9984,19193,2507611748,90452328,29027172,2026-03-20T02:08:02Z,239,0,34,16
9988
  CC-MAIN-2026-08,9985,19367,2531391271,94995010,30776832,2026-03-20T02:10:32Z,166,0,38,14
9989
  CC-MAIN-2026-08,9986,19589,2541080212,94152803,29609822,2026-03-20T03:23:17Z,174,0,25,50