tamnd commited on
Commit
1598165
·
verified ·
1 Parent(s): 17c0775

Publish shard CC-MAIN-2026-08/00518

Browse files
Files changed (3) hide show
  1. README.md +11 -11
  2. data/CC-MAIN-2026-08/00518.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 **17,244,271 documents across 891 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 891 files of CC-MAIN-2026-08 (17,244,271 pages total), projected to the full crawl of 100,000 WARC files.
168
 
169
- | Stage | 891 files (measured) | 100,000 files (projected) | Reduction |
170
  |---|---|---|---|
171
- | Raw WARC (.warc.gz, downloaded) | ~722.2 GB | ~83 TB | — |
172
  | HTML extracted (uncompressed) | 2.1 TB | ~295 TB | — |
173
- | Packed markdown WARC (.md.warc.gz) | ~36.5 GB | ~3.7 TB | **-98.3%** vs HTML |
174
  | Final Parquet (Zstd level 19) | 24.7 GB | ~2.9 TB | **-32.4%** 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 2.1 TB of uncompressed HTML down to 77.7 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 68.2%.
177
 
178
- End to end: ~722.2 GB of raw gzipped WARCs becomes **24.7 GB of Parquet** — a **96.6% total reduction** — containing 17,244,271 clean markdown documents.
179
 
180
  ### Processing Times
181
 
182
- Pipeline timings across 891 shards of CC-MAIN-2026-08:
183
 
184
  ```
185
- Download (raw WARC) ████████████████████████ total 54h 44m 16s avg 3m 41s
186
  Convert (HTML → MD) █░░░░░░░░░░░░░░░░░░░░░░░ total 4h 30m 54s avg 18s
187
- Export (Parquet) █████░░░░░░░░░░░░░░░░░░░ total 12h 27m 36s avg 50s
188
- Publish (HuggingFace) ██░░░░░░░░░░░░░░░░░░░░░░ total 6h 27m 9s avg 26s
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 **17,263,881 documents across 892 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 892 files of CC-MAIN-2026-08 (17,263,881 pages total), projected to the full crawl of 100,000 WARC files.
168
 
169
+ | Stage | 892 files (measured) | 100,000 files (projected) | Reduction |
170
  |---|---|---|---|
171
+ | Raw WARC (.warc.gz, downloaded) | ~723.0 GB | ~83 TB | — |
172
  | HTML extracted (uncompressed) | 2.1 TB | ~295 TB | — |
173
+ | Packed markdown WARC (.md.warc.gz) | ~36.6 GB | ~3.7 TB | **-98.3%** vs HTML |
174
  | Final Parquet (Zstd level 19) | 24.7 GB | ~2.9 TB | **-32.4%** 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 2.1 TB of uncompressed HTML down to 77.8 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 68.2%.
177
 
178
+ End to end: ~723.0 GB of raw gzipped WARCs becomes **24.7 GB of Parquet** — a **96.6% total reduction** — containing 17,263,881 clean markdown documents.
179
 
180
  ### Processing Times
181
 
182
+ Pipeline timings across 892 shards of CC-MAIN-2026-08:
183
 
184
  ```
185
+ Download (raw WARC) ████████████████████████ total 54h 52m 18s avg 3m 41s
186
  Convert (HTML → MD) █░░░░░░░░░░░░░░░░░░░░░░░ total 4h 30m 54s avg 18s
187
+ Export (Parquet) █████░░░░░░░░░░░░░░░░░░░ total 12h 28m 49s avg 50s
188
+ Publish (HuggingFace) ██░░░░░░░░░░░░░░░░░░░░░░ total 6h 27m 26s avg 26s
189
  ```
190
 
191
  ### Dataset Charts
data/CC-MAIN-2026-08/00518.parquet ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:c34dede11a944ff76c9b341a83e9530be2a3015b25bba9fbba8f32b9166ea2a6
3
+ size 30705729
stats.csv CHANGED
@@ -281,7 +281,7 @@ CC-MAIN-2026-08,342,19280,2567257308,93681727,29660197,2026-03-13T19:36:52Z,0,35
281
  CC-MAIN-2026-08,343,19262,2539673272,90965417,28370002,2026-03-13T19:48:56Z,643,0,77,35
282
  CC-MAIN-2026-08,344,19491,2610007404,98525224,31295176,2026-03-13T19:58:19Z,466,0,65,22
283
  CC-MAIN-2026-08,345,19638,2571646305,92700336,29370025,2026-03-13T20:08:56Z,582,0,85,46
284
- CC-MAIN-2026-08,346,19473,2585582152,94779939,29804017,2026-03-13T20:18:11Z,477,0,73,0
285
  CC-MAIN-2026-08,401,19441,2586676962,94334130,29618260,2026-03-13T11:41:00Z,0,0,33,0
286
  CC-MAIN-2026-08,402,19446,2552785517,96654791,30560442,2026-03-13T11:49:21Z,0,443,57,0
287
  CC-MAIN-2026-08,403,19320,2579048144,91406292,29341198,2026-03-13T11:56:46Z,0,414,29,0
@@ -345,6 +345,7 @@ CC-MAIN-2026-08,514,19282,2577246804,93966169,29763725,2026-03-13T19:38:27Z,542,
345
  CC-MAIN-2026-08,515,19316,2560287042,96839519,29930731,2026-03-13T19:49:44Z,600,0,77,14
346
  CC-MAIN-2026-08,516,19528,2607927055,95622899,30793330,2026-03-13T19:59:23Z,469,0,68,29
347
  CC-MAIN-2026-08,517,19333,2578621012,89099863,28880437,2026-03-13T20:10:04Z,547,0,89,17
 
348
  CC-MAIN-2026-08,551,19551,2609034951,89980913,29167227,2026-03-13T06:42:51Z,280,0,20,176
349
  CC-MAIN-2026-08,552,19402,2594817630,94092761,30309254,2026-03-13T06:48:30Z,211,0,46,80
350
  CC-MAIN-2026-08,553,19856,2617133708,99106486,30365588,2026-03-13T06:52:51Z,116,0,23,120
 
281
  CC-MAIN-2026-08,343,19262,2539673272,90965417,28370002,2026-03-13T19:48:56Z,643,0,77,35
282
  CC-MAIN-2026-08,344,19491,2610007404,98525224,31295176,2026-03-13T19:58:19Z,466,0,65,22
283
  CC-MAIN-2026-08,345,19638,2571646305,92700336,29370025,2026-03-13T20:08:56Z,582,0,85,46
284
+ CC-MAIN-2026-08,346,19473,2585582152,94779939,29804017,2026-03-13T20:18:11Z,477,0,73,17
285
  CC-MAIN-2026-08,401,19441,2586676962,94334130,29618260,2026-03-13T11:41:00Z,0,0,33,0
286
  CC-MAIN-2026-08,402,19446,2552785517,96654791,30560442,2026-03-13T11:49:21Z,0,443,57,0
287
  CC-MAIN-2026-08,403,19320,2579048144,91406292,29341198,2026-03-13T11:56:46Z,0,414,29,0
 
345
  CC-MAIN-2026-08,515,19316,2560287042,96839519,29930731,2026-03-13T19:49:44Z,600,0,77,14
346
  CC-MAIN-2026-08,516,19528,2607927055,95622899,30793330,2026-03-13T19:59:23Z,469,0,68,29
347
  CC-MAIN-2026-08,517,19333,2578621012,89099863,28880437,2026-03-13T20:10:04Z,547,0,89,17
348
+ CC-MAIN-2026-08,518,19610,2602806405,95330083,30705729,2026-03-13T20:18:36Z,482,0,73,0
349
  CC-MAIN-2026-08,551,19551,2609034951,89980913,29167227,2026-03-13T06:42:51Z,280,0,20,176
350
  CC-MAIN-2026-08,552,19402,2594817630,94092761,30309254,2026-03-13T06:48:30Z,211,0,46,80
351
  CC-MAIN-2026-08,553,19856,2617133708,99106486,30365588,2026-03-13T06:52:51Z,116,0,23,120