tamnd commited on
Commit
03d6e8d
·
verified ·
1 Parent(s): 9f4d6dd

Publish shard CC-MAIN-2026-08/02584

Browse files
Files changed (3) hide show
  1. README.md +10 -10
  2. data/CC-MAIN-2026-08/02584.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 **14,578,159 documents across 754 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 754 files of CC-MAIN-2026-08 (14,578,159 pages total), projected to the full crawl of 100,000 WARC files.
168
 
169
- | Stage | 754 files (measured) | 100,000 files (projected) | Reduction |
170
  |---|---|---|---|
171
- | Raw WARC (.warc.gz, downloaded) | ~611.2 GB | ~83 TB | — |
172
  | HTML extracted (uncompressed) | 1.8 TB | ~295 TB | — |
173
  | Packed markdown WARC (.md.warc.gz) | ~30.9 GB | ~3.7 TB | **-98.3%** vs HTML |
174
  | Final Parquet (Zstd level 19) | 20.9 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 1.8 TB of uncompressed HTML down to 65.7 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: ~611.2 GB of raw gzipped WARCs becomes **20.9 GB of Parquet** — a **96.6% total reduction** — containing 14,578,159 clean markdown documents.
179
 
180
  ### Processing Times
181
 
182
- Pipeline timings across 754 shards of CC-MAIN-2026-08:
183
 
184
  ```
185
- Download (raw WARC) ████████████████████████ total 42h 6m 9s avg 3m 21s
186
  Convert (HTML → MD) █░░░░░░░░░░░░░░░░░░░░░░░ total 1h 50m 33s avg 8s
187
- Export (Parquet) █████░░░░░░░░░░░░░░░░░░░ total 9h 7m 32s avg 43s
188
- Publish (HuggingFace) ███░░░░░░░░░░░░░░░░░░░░░ total 5h 36m 10s 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 **14,597,643 documents across 755 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 755 files of CC-MAIN-2026-08 (14,597,643 pages total), projected to the full crawl of 100,000 WARC files.
168
 
169
+ | Stage | 755 files (measured) | 100,000 files (projected) | Reduction |
170
  |---|---|---|---|
171
+ | Raw WARC (.warc.gz, downloaded) | ~612.0 GB | ~83 TB | — |
172
  | HTML extracted (uncompressed) | 1.8 TB | ~295 TB | — |
173
  | Packed markdown WARC (.md.warc.gz) | ~30.9 GB | ~3.7 TB | **-98.3%** vs HTML |
174
  | Final Parquet (Zstd level 19) | 20.9 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 1.8 TB of uncompressed HTML down to 65.8 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: ~612.0 GB of raw gzipped WARCs becomes **20.9 GB of Parquet** — a **96.6% total reduction** — containing 14,597,643 clean markdown documents.
179
 
180
  ### Processing Times
181
 
182
+ Pipeline timings across 755 shards of CC-MAIN-2026-08:
183
 
184
  ```
185
+ Download (raw WARC) ████████████████████████ total 42h 26m 15s avg 3m 22s
186
  Convert (HTML → MD) █░░░░░░░░░░░░░░░░░░░░░░░ total 1h 50m 33s avg 8s
187
+ Export (Parquet) █████░░░░░░░░░░░░░░░░░░░ total 9h 8m 28s avg 43s
188
+ Publish (HuggingFace) ███░░░░░░░░░░░░░░░░░░░░░ total 5h 36m 33s avg 26s
189
  ```
190
 
191
  ### Dataset Charts
data/CC-MAIN-2026-08/02584.parquet ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:2e0deeca3d511b1f4e241ab8f4d3ea6159baa2022eff253b71d905000f42943a
3
+ size 30756815
stats.csv CHANGED
@@ -699,6 +699,7 @@ CC-MAIN-2026-08,2580,19296,2524393224,91578828,29472992,2026-03-13T15:21:56Z,318
699
  CC-MAIN-2026-08,2581,19471,2567634976,95674819,29606440,2026-03-13T15:26:19Z,441,0,39,19
700
  CC-MAIN-2026-08,2582,19563,2556513838,98156888,31148205,2026-03-13T15:37:54Z,617,0,49,9
701
  CC-MAIN-2026-08,2583,19288,2578146937,96424818,29805976,2026-03-13T16:00:54Z,1319,0,71,13
 
702
  CC-MAIN-2026-08,2801,19800,2573332995,93935647,29781057,2026-03-13T12:36:24Z,480,0,24,0
703
  CC-MAIN-2026-08,2802,19377,2552826046,96926255,30379332,2026-03-13T12:44:04Z,402,0,56,0
704
  CC-MAIN-2026-08,2810,19479,2567872535,94762671,29742907,2026-03-13T14:03:56Z,0,0,0,4
@@ -719,7 +720,7 @@ CC-MAIN-2026-08,2825,19533,2582104096,95896575,30321794,2026-03-13T15:52:41Z,305
719
  CC-MAIN-2026-08,2826,19519,2588031935,96747984,30634522,2026-03-13T16:00:26Z,393,0,80,23
720
  CC-MAIN-2026-08,2827,19399,2545980437,91425550,29777608,2026-03-13T16:08:56Z,467,0,35,13
721
  CC-MAIN-2026-08,2828,19552,2560245417,89820454,28768898,2026-03-13T16:12:41Z,193,0,39,11
722
- CC-MAIN-2026-08,2829,19457,2574659951,95090234,29784169,2026-03-13T16:20:57Z,386,0,98,0
723
  CC-MAIN-2026-08,3051,19591,2590387140,91599959,29213641,2026-03-13T12:27:59Z,0,0,0,5
724
  CC-MAIN-2026-08,3052,19289,2581535769,95779278,29412368,2026-03-13T12:27:59Z,0,0,0,5
725
  CC-MAIN-2026-08,3053,19208,2553598556,93105627,29687407,2026-03-13T12:27:59Z,0,0,0,5
 
699
  CC-MAIN-2026-08,2581,19471,2567634976,95674819,29606440,2026-03-13T15:26:19Z,441,0,39,19
700
  CC-MAIN-2026-08,2582,19563,2556513838,98156888,31148205,2026-03-13T15:37:54Z,617,0,49,9
701
  CC-MAIN-2026-08,2583,19288,2578146937,96424818,29805976,2026-03-13T16:00:54Z,1319,0,71,13
702
+ CC-MAIN-2026-08,2584,19484,2553112375,97091387,30756815,2026-03-13T16:21:41Z,1206,0,56,0
703
  CC-MAIN-2026-08,2801,19800,2573332995,93935647,29781057,2026-03-13T12:36:24Z,480,0,24,0
704
  CC-MAIN-2026-08,2802,19377,2552826046,96926255,30379332,2026-03-13T12:44:04Z,402,0,56,0
705
  CC-MAIN-2026-08,2810,19479,2567872535,94762671,29742907,2026-03-13T14:03:56Z,0,0,0,4
 
720
  CC-MAIN-2026-08,2826,19519,2588031935,96747984,30634522,2026-03-13T16:00:26Z,393,0,80,23
721
  CC-MAIN-2026-08,2827,19399,2545980437,91425550,29777608,2026-03-13T16:08:56Z,467,0,35,13
722
  CC-MAIN-2026-08,2828,19552,2560245417,89820454,28768898,2026-03-13T16:12:41Z,193,0,39,11
723
+ CC-MAIN-2026-08,2829,19457,2574659951,95090234,29784169,2026-03-13T16:20:57Z,386,0,98,23
724
  CC-MAIN-2026-08,3051,19591,2590387140,91599959,29213641,2026-03-13T12:27:59Z,0,0,0,5
725
  CC-MAIN-2026-08,3052,19289,2581535769,95779278,29412368,2026-03-13T12:27:59Z,0,0,0,5
726
  CC-MAIN-2026-08,3053,19208,2553598556,93105627,29687407,2026-03-13T12:27:59Z,0,0,0,5