tamnd commited on
Commit
28b68e4
·
verified ·
1 Parent(s): a362574

Publish shard CC-MAIN-2026-08/00750

Browse files
Files changed (3) hide show
  1. README.md +9 -9
  2. data/CC-MAIN-2026-08/00750.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 **15,183,332 documents across 785 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 785 files of CC-MAIN-2026-08 (15,183,332 pages total), projected to the full crawl of 100,000 WARC files.
168
 
169
- | Stage | 785 files (measured) | 100,000 files (projected) | Reduction |
170
  |---|---|---|---|
171
- | Raw WARC (.warc.gz, downloaded) | ~636.3 GB | ~83 TB | — |
172
  | HTML extracted (uncompressed) | 1.8 TB | ~295 TB | — |
173
  | Packed markdown WARC (.md.warc.gz) | ~32.2 GB | ~3.7 TB | **-98.3%** vs HTML |
174
  | Final Parquet (Zstd level 19) | 21.8 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 68.5 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: ~636.3 GB of raw gzipped WARCs becomes **21.8 GB of Parquet** — a **96.6% total reduction** — containing 15,183,332 clean markdown documents.
179
 
180
  ### Processing Times
181
 
182
- Pipeline timings across 785 shards of CC-MAIN-2026-08:
183
 
184
  ```
185
- Download (raw WARC) ████████████████████████ total 44h 35m 49s avg 3m 24s
186
  Convert (HTML → MD) █░░░░░░░░░░░░░░░░░░░░░░░ total 1h 56m 12s avg 8s
187
- Export (Parquet) █████░░░░░░░░░░░░░░░░░░░ total 9h 30m 4s avg 43s
188
- Publish (HuggingFace) ███░░░░░░░░░░░░░░░░░░░░░ total 5h 51m 34s 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 **15,202,803 documents across 786 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 786 files of CC-MAIN-2026-08 (15,202,803 pages total), projected to the full crawl of 100,000 WARC files.
168
 
169
+ | Stage | 786 files (measured) | 100,000 files (projected) | Reduction |
170
  |---|---|---|---|
171
+ | Raw WARC (.warc.gz, downloaded) | ~637.1 GB | ~83 TB | — |
172
  | HTML extracted (uncompressed) | 1.8 TB | ~295 TB | — |
173
  | Packed markdown WARC (.md.warc.gz) | ~32.2 GB | ~3.7 TB | **-98.3%** vs HTML |
174
  | Final Parquet (Zstd level 19) | 21.8 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 68.5 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: ~637.1 GB of raw gzipped WARCs becomes **21.8 GB of Parquet** — a **96.6% total reduction** — containing 15,202,803 clean markdown documents.
179
 
180
  ### Processing Times
181
 
182
+ Pipeline timings across 786 shards of CC-MAIN-2026-08:
183
 
184
  ```
185
+ Download (raw WARC) ████████████████████████ total 44h 43m 4s avg 3m 24s
186
  Convert (HTML → MD) █░░░░░░░░░░░░░░░░░░░░░░░ total 1h 56m 12s avg 8s
187
+ Export (Parquet) █████░░░░░░░░░░░░░░░░░░░ total 9h 31m 14s avg 43s
188
+ Publish (HuggingFace) ███░░░░░░░░░░░░░░░░░░░░░ total 5h 52m 0s avg 26s
189
  ```
190
 
191
  ### Dataset Charts
data/CC-MAIN-2026-08/00750.parquet ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:f476a28ea82736fa79ddf79071116137fee1fed60551fcdb357ff5c588376a0b
3
+ size 29734352
stats.csv CHANGED
@@ -384,6 +384,7 @@ CC-MAIN-2026-08,738,19381,2580510880,97427741,31334407,2026-03-13T11:42:32Z,236,
384
  CC-MAIN-2026-08,739,19722,2589544973,95675374,30075683,2026-03-13T11:50:09Z,402,0,54,9
385
  CC-MAIN-2026-08,740,19301,2556616688,93201201,30300647,2026-03-13T11:55:56Z,310,0,35,9
386
  CC-MAIN-2026-08,741,19382,2570241665,94898311,30073507,2026-03-13T12:01:53Z,223,0,33,0
 
387
  CC-MAIN-2026-08,851,19575,2612873485,97000364,30911294,2026-03-13T06:41:39Z,163,0,37,204
388
  CC-MAIN-2026-08,852,19334,2546015208,94959847,30205044,2026-03-13T06:45:52Z,183,0,57,11
389
  CC-MAIN-2026-08,853,19694,2567973975,95901932,30303494,2026-03-13T06:50:12Z,150,0,28,80
@@ -517,7 +518,7 @@ CC-MAIN-2026-08,1204,19264,2538327084,94846804,29759525,2026-03-13T11:50:12Z,455
517
  CC-MAIN-2026-08,1205,19364,2588047993,93773809,29550655,2026-03-13T11:54:07Z,165,0,69,0
518
  CC-MAIN-2026-08,1206,19390,2574716406,95446241,30104933,2026-03-13T11:57:07Z,152,0,27,0
519
  CC-MAIN-2026-08,1207,19457,2587304186,94642420,29428000,2026-03-13T12:01:50Z,205,0,76,0
520
- CC-MAIN-2026-08,1250,19570,2579186072,96548314,30182713,2026-03-13T17:40:42Z,339,0,97,0
521
  CC-MAIN-2026-08,1301,19417,2601765043,93758124,29790521,2026-03-13T11:40:27Z,0,0,0,0
522
  CC-MAIN-2026-08,1302,19471,2592470194,95488689,29747494,2026-03-13T11:40:27Z,0,0,0,0
523
  CC-MAIN-2026-08,1303,19681,2609544680,95027524,30250671,2026-03-13T11:40:27Z,0,0,0,0
 
384
  CC-MAIN-2026-08,739,19722,2589544973,95675374,30075683,2026-03-13T11:50:09Z,402,0,54,9
385
  CC-MAIN-2026-08,740,19301,2556616688,93201201,30300647,2026-03-13T11:55:56Z,310,0,35,9
386
  CC-MAIN-2026-08,741,19382,2570241665,94898311,30073507,2026-03-13T12:01:53Z,223,0,33,0
387
+ CC-MAIN-2026-08,750,19471,2557384086,90848659,29734352,2026-03-13T17:41:16Z,435,0,70,0
388
  CC-MAIN-2026-08,851,19575,2612873485,97000364,30911294,2026-03-13T06:41:39Z,163,0,37,204
389
  CC-MAIN-2026-08,852,19334,2546015208,94959847,30205044,2026-03-13T06:45:52Z,183,0,57,11
390
  CC-MAIN-2026-08,853,19694,2567973975,95901932,30303494,2026-03-13T06:50:12Z,150,0,28,80
 
518
  CC-MAIN-2026-08,1205,19364,2588047993,93773809,29550655,2026-03-13T11:54:07Z,165,0,69,0
519
  CC-MAIN-2026-08,1206,19390,2574716406,95446241,30104933,2026-03-13T11:57:07Z,152,0,27,0
520
  CC-MAIN-2026-08,1207,19457,2587304186,94642420,29428000,2026-03-13T12:01:50Z,205,0,76,0
521
+ CC-MAIN-2026-08,1250,19570,2579186072,96548314,30182713,2026-03-13T17:40:42Z,339,0,97,26
522
  CC-MAIN-2026-08,1301,19417,2601765043,93758124,29790521,2026-03-13T11:40:27Z,0,0,0,0
523
  CC-MAIN-2026-08,1302,19471,2592470194,95488689,29747494,2026-03-13T11:40:27Z,0,0,0,0
524
  CC-MAIN-2026-08,1303,19681,2609544680,95027524,30250671,2026-03-13T11:40:27Z,0,0,0,0