tamnd commited on
Commit
49d4e8f
·
verified ·
1 Parent(s): 10618df

Publish shard CC-MAIN-2026-08/00252

Browse files
Files changed (3) hide show
  1. README.md +13 -13
  2. data/CC-MAIN-2026-08/00252.parquet +3 -0
  3. stats.csv +3 -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 **990,718 documents across 50 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
 
@@ -166,28 +166,28 @@ Empty conversions (pages where trafilatura could not extract meaningful content)
166
 
167
  ### Compression Ratios
168
 
169
- Numbers below are actual measurements summed across all 50 files of CC-MAIN-2026-08 (990,718 pages total), projected to the full crawl of 100,000 WARC files.
170
 
171
- | Stage | 50 files (measured) | 100,000 files (projected) | Reduction |
172
  |---|---|---|---|
173
- | Raw WARC (.warc.gz, downloaded) | ~40.5 GB | ~83 TB | — |
174
- | HTML extracted (uncompressed) | 123.8 GB | ~295 TB | — |
175
- | Packed markdown WARC (.md.warc.gz) | ~2.0 GB | ~3.7 TB | **-98.4%** vs HTML |
176
  | Final Parquet (Zstd level 19) | 1.4 GB | ~2.9 TB | **-31.4%** vs packed WARC |
177
 
178
- The big win is the HTML → Markdown step: trafilatura strips all tags, scripts, styles, navigation, and ads, keeping only the main content. This cuts 123.8 GB of uncompressed HTML down to 4.2 GB of markdown — a **98.4% reduction** — before any file-level compression is applied. Parquet with Zstd level 19 then compresses the markdown a further 67.7%.
179
 
180
- End to end: ~40.5 GB of raw gzipped WARCs becomes **1.4 GB of Parquet** — a **96.6% total reduction** — containing 990,718 clean markdown documents.
181
 
182
  ### Processing Times
183
 
184
- Pipeline timings across 50 shards of CC-MAIN-2026-08:
185
 
186
  ```
187
- Download (raw WARC) ████████████████████████ total 36m 16s avg 43s
188
- Convert (HTML → MD) ██████████░░░░░░░░░░░░░░ total 16m 2s avg 19s
189
- Export (Parquet) ██████░░░░░░░░░░░░░░░░░░ total 10m 29s avg 12s
190
- Publish (HuggingFace) ██████████████░░░░░░░░░░ total 21m 36s avg 25s
191
  ```
192
 
193
  ### 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 **1,029,469 documents across 52 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
 
 
166
 
167
  ### Compression Ratios
168
 
169
+ Numbers below are actual measurements summed across all 52 files of CC-MAIN-2026-08 (1,029,469 pages total), projected to the full crawl of 100,000 WARC files.
170
 
171
+ | Stage | 52 files (measured) | 100,000 files (projected) | Reduction |
172
  |---|---|---|---|
173
+ | Raw WARC (.warc.gz, downloaded) | ~42.1 GB | ~83 TB | — |
174
+ | HTML extracted (uncompressed) | 128.6 GB | ~295 TB | — |
175
+ | Packed markdown WARC (.md.warc.gz) | ~2.1 GB | ~3.7 TB | **-98.4%** vs HTML |
176
  | Final Parquet (Zstd level 19) | 1.4 GB | ~2.9 TB | **-31.4%** vs packed WARC |
177
 
178
+ The big win is the HTML → Markdown step: trafilatura strips all tags, scripts, styles, navigation, and ads, keeping only the main content. This cuts 128.6 GB of uncompressed HTML down to 4.4 GB of markdown — a **98.4% reduction** — before any file-level compression is applied. Parquet with Zstd level 19 then compresses the markdown a further 67.8%.
179
 
180
+ End to end: ~42.1 GB of raw gzipped WARCs becomes **1.4 GB of Parquet** — a **96.6% total reduction** — containing 1,029,469 clean markdown documents.
181
 
182
  ### Processing Times
183
 
184
+ Pipeline timings across 52 shards of CC-MAIN-2026-08:
185
 
186
  ```
187
+ Download (raw WARC) ████████████████████████ total 43m 21s avg 50s
188
+ Convert (HTML → MD) ████████░░░░░░░░░░░░░░░░ total 16m 2s avg 18s
189
+ Export (Parquet) ██████░░░░░░░░░░░░░░░░░░ total 12m 4s avg 13s
190
+ Publish (HuggingFace) ████████████░░░░░░░░░░░░ total 21m 54s avg 25s
191
  ```
192
 
193
  ### Dataset Charts
data/CC-MAIN-2026-08/00252.parquet ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:c3bb148b34d893db211b1708049c9ffa5ccee67fd8ac58b202aa85b14fcab5c7
3
+ size 31104034
stats.csv CHANGED
@@ -38,10 +38,12 @@ CC-MAIN-2026-08,35,19398,2565260259,94010975,30284205,2026-03-13T06:06:37Z,0,72,
38
  CC-MAIN-2026-08,36,19367,2577531843,94004962,29843828,2026-03-13T06:08:12Z,0,69,18,2
39
  CC-MAIN-2026-08,37,19314,2558501740,95109133,29807147,2026-03-13T06:37:54Z,0,63,19,96
40
  CC-MAIN-2026-08,38,19372,2618600358,93097292,29730152,2026-03-13T06:42:06Z,131,0,20,99
41
- CC-MAIN-2026-08,39,19660,2620924368,92727175,29965496,2026-03-13T06:45:58Z,169,0,61,0
42
  CC-MAIN-2026-08,101,19267,2514449658,96925434,30875010,2026-03-13T06:40:05Z,179,0,50,81
43
  CC-MAIN-2026-08,102,19438,2568537422,97402434,31137610,2026-03-13T06:42:52Z,93,0,20,52
 
44
  CC-MAIN-2026-08,251,19610,2594898345,97558492,31917087,2026-03-13T06:42:12Z,277,0,23,136
 
45
  CC-MAIN-2026-08,401,19517,2605589961,95609438,30404753,2026-03-13T06:41:08Z,218,0,47,107
46
  CC-MAIN-2026-08,402,19446,2552785517,96654791,30588984,2026-03-13T06:43:58Z,98,0,26,44
47
  CC-MAIN-2026-08,551,19551,2609034951,89980913,29167227,2026-03-13T06:42:51Z,280,0,20,176
 
38
  CC-MAIN-2026-08,36,19367,2577531843,94004962,29843828,2026-03-13T06:08:12Z,0,69,18,2
39
  CC-MAIN-2026-08,37,19314,2558501740,95109133,29807147,2026-03-13T06:37:54Z,0,63,19,96
40
  CC-MAIN-2026-08,38,19372,2618600358,93097292,29730152,2026-03-13T06:42:06Z,131,0,20,99
41
+ CC-MAIN-2026-08,39,19660,2620924368,92727175,29965496,2026-03-13T06:46:16Z,169,0,61,18
42
  CC-MAIN-2026-08,101,19267,2514449658,96925434,30875010,2026-03-13T06:40:05Z,179,0,50,81
43
  CC-MAIN-2026-08,102,19438,2568537422,97402434,31137610,2026-03-13T06:42:52Z,93,0,20,52
44
+ CC-MAIN-2026-08,103,19511,2593312947,96164480,30296223,2026-03-13T06:46:46Z,187,0,46,0
45
  CC-MAIN-2026-08,251,19610,2594898345,97558492,31917087,2026-03-13T06:42:12Z,277,0,23,136
46
+ CC-MAIN-2026-08,252,19240,2525928701,98163046,31104034,2026-03-13T06:47:00Z,238,0,49,0
47
  CC-MAIN-2026-08,401,19517,2605589961,95609438,30404753,2026-03-13T06:41:08Z,218,0,47,107
48
  CC-MAIN-2026-08,402,19446,2552785517,96654791,30588984,2026-03-13T06:43:58Z,98,0,26,44
49
  CC-MAIN-2026-08,551,19551,2609034951,89980913,29167227,2026-03-13T06:42:51Z,280,0,20,176