tamnd commited on
Commit
7d3daf6
·
verified ·
1 Parent(s): 323aa75

Publish shard CC-MAIN-2026-08/05429

Browse files
Files changed (3) hide show
  1. README.md +9 -9
  2. data/CC-MAIN-2026-08/05429.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 **49,126,247 documents across 2535 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 2535 files of CC-MAIN-2026-08 (49,126,247 pages total), projected to the full crawl of 100,000 WARC files.
168
 
169
- | Stage | 2535 files (measured) | 100,000 files (projected) | Reduction |
170
  |---|---|---|---|
171
  | Raw WARC (.warc.gz, downloaded) | ~2.0 TB | ~83 TB | — |
172
  | HTML extracted (uncompressed) | 5.9 TB | ~295 TB | — |
173
  | Packed markdown WARC (.md.warc.gz) | ~103.9 GB | ~3.7 TB | **-98.3%** vs HTML |
174
  | Final Parquet (Zstd level 19) | 70.1 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 5.9 TB of uncompressed HTML down to 221.0 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: ~2.0 TB of raw gzipped WARCs becomes **70.1 GB of Parquet** — a **96.6% total reduction** — containing 49,126,247 clean markdown documents.
179
 
180
  ### Processing Times
181
 
182
- Pipeline timings across 2535 shards of CC-MAIN-2026-08:
183
 
184
  ```
185
- Download (raw WARC) ████████████████████████ total 219h 0m 56s avg 5m 11s
186
  Convert (HTML → MD) █░░░░░░░░░░░░░░░░░░░░░░░ total 12h 57m 19s avg 18s
187
- Export (Parquet) ██████░░░░░░░░░░░░░░░░░░ total 56h 34m 58s avg 1m 20s
188
- Publish (HuggingFace) ██░░░░░░░░░░░░░░░░░░░░░░ total 19h 15m 31s avg 27s
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 **49,145,795 documents across 2536 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 2536 files of CC-MAIN-2026-08 (49,145,795 pages total), projected to the full crawl of 100,000 WARC files.
168
 
169
+ | Stage | 2536 files (measured) | 100,000 files (projected) | Reduction |
170
  |---|---|---|---|
171
  | Raw WARC (.warc.gz, downloaded) | ~2.0 TB | ~83 TB | — |
172
  | HTML extracted (uncompressed) | 5.9 TB | ~295 TB | — |
173
  | Packed markdown WARC (.md.warc.gz) | ~103.9 GB | ~3.7 TB | **-98.3%** vs HTML |
174
  | Final Parquet (Zstd level 19) | 70.1 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 5.9 TB of uncompressed HTML down to 221.1 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: ~2.0 TB of raw gzipped WARCs becomes **70.1 GB of Parquet** — a **96.6% total reduction** — containing 49,145,795 clean markdown documents.
179
 
180
  ### Processing Times
181
 
182
+ Pipeline timings across 2536 shards of CC-MAIN-2026-08:
183
 
184
  ```
185
+ Download (raw WARC) ████████████████████████ total 219h 5m 8s avg 5m 11s
186
  Convert (HTML → MD) █░░░░░░░░░░░░░░░░░░░░░░░ total 12h 57m 19s avg 18s
187
+ Export (Parquet) ██████░░░░░░░░░░░░░░░░░░ total 56h 35m 36s avg 1m 20s
188
+ Publish (HuggingFace) ██░░░░░░░░░░░░░░░░░░░░░░ total 19h 16m 22s avg 27s
189
  ```
190
 
191
  ### Dataset Charts
data/CC-MAIN-2026-08/05429.parquet ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:4e77296c577f7cfc57e1740183e6bee0d163790b7043fe858267f28e651292a8
3
+ size 30323087
stats.csv CHANGED
@@ -1866,6 +1866,7 @@ CC-MAIN-2026-08,5425,19429,2576314244,94307917,29745400,2026-03-14T17:00:36Z,260
1866
  CC-MAIN-2026-08,5426,19324,2584747862,95302503,29739381,2026-03-14T17:05:33Z,266,0,70,37
1867
  CC-MAIN-2026-08,5427,19592,2607413413,94493352,30576986,2026-03-14T17:10:46Z,218,0,84,26
1868
  CC-MAIN-2026-08,5428,19503,2578508339,91813650,29390477,2026-03-14T17:15:33Z,235,0,64,15
 
1869
  CC-MAIN-2026-08,5492,19488,2577487663,95676086,30636918,2026-03-14T01:06:35Z,267,0,18,18
1870
  CC-MAIN-2026-08,5493,19631,2592234390,91001440,28842612,2026-03-14T01:19:12Z,181,0,28,0
1871
  CC-MAIN-2026-08,5494,19345,2580777219,94618705,30365899,2026-03-14T01:19:12Z,155,0,49,0
@@ -2374,7 +2375,7 @@ CC-MAIN-2026-08,6158,19638,2623537896,92701477,29359245,2026-03-14T16:59:11Z,232
2374
  CC-MAIN-2026-08,6159,19291,2561228706,89317858,28275692,2026-03-14T17:03:56Z,214,0,81,13
2375
  CC-MAIN-2026-08,6160,19407,2541135298,89309847,28260533,2026-03-14T17:09:09Z,193,0,72,14
2376
  CC-MAIN-2026-08,6161,19466,2582668574,91714130,29766122,2026-03-14T17:14:13Z,253,0,77,14
2377
- CC-MAIN-2026-08,6162,19501,2576375344,90765634,29197495,2026-03-14T17:19:24Z,242,0,55,0
2378
  CC-MAIN-2026-08,6250,19502,2594302344,89411296,28384540,2026-03-13T17:51:21Z,406,0,120,0
2379
  CC-MAIN-2026-08,6251,19394,2555453090,92793633,28885378,2026-03-14T01:06:36Z,709,0,233,18
2380
  CC-MAIN-2026-08,6252,19389,2566928631,91567909,28780284,2026-03-14T01:06:36Z,515,0,174,18
 
1866
  CC-MAIN-2026-08,5426,19324,2584747862,95302503,29739381,2026-03-14T17:05:33Z,266,0,70,37
1867
  CC-MAIN-2026-08,5427,19592,2607413413,94493352,30576986,2026-03-14T17:10:46Z,218,0,84,26
1868
  CC-MAIN-2026-08,5428,19503,2578508339,91813650,29390477,2026-03-14T17:15:33Z,235,0,64,15
1869
+ CC-MAIN-2026-08,5429,19548,2594907917,96706876,30323087,2026-03-14T17:20:17Z,252,0,38,0
1870
  CC-MAIN-2026-08,5492,19488,2577487663,95676086,30636918,2026-03-14T01:06:35Z,267,0,18,18
1871
  CC-MAIN-2026-08,5493,19631,2592234390,91001440,28842612,2026-03-14T01:19:12Z,181,0,28,0
1872
  CC-MAIN-2026-08,5494,19345,2580777219,94618705,30365899,2026-03-14T01:19:12Z,155,0,49,0
 
2375
  CC-MAIN-2026-08,6159,19291,2561228706,89317858,28275692,2026-03-14T17:03:56Z,214,0,81,13
2376
  CC-MAIN-2026-08,6160,19407,2541135298,89309847,28260533,2026-03-14T17:09:09Z,193,0,72,14
2377
  CC-MAIN-2026-08,6161,19466,2582668574,91714130,29766122,2026-03-14T17:14:13Z,253,0,77,14
2378
+ CC-MAIN-2026-08,6162,19501,2576375344,90765634,29197495,2026-03-14T17:19:24Z,242,0,55,51
2379
  CC-MAIN-2026-08,6250,19502,2594302344,89411296,28384540,2026-03-13T17:51:21Z,406,0,120,0
2380
  CC-MAIN-2026-08,6251,19394,2555453090,92793633,28885378,2026-03-14T01:06:36Z,709,0,233,18
2381
  CC-MAIN-2026-08,6252,19389,2566928631,91567909,28780284,2026-03-14T01:06:36Z,515,0,174,18