tamnd commited on
Commit
0a2ab98
·
verified ·
1 Parent(s): 3c2619a

Publish shard CC-MAIN-2026-08/00336

Browse files
Files changed (3) hide show
  1. README.md +11 -11
  2. data/CC-MAIN-2026-08/00336.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 **16,058,095 documents across 830 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 830 files of CC-MAIN-2026-08 (16,058,095 pages total), projected to the full crawl of 100,000 WARC files.
168
 
169
- | Stage | 830 files (measured) | 100,000 files (projected) | Reduction |
170
  |---|---|---|---|
171
- | Raw WARC (.warc.gz, downloaded) | ~672.8 GB | ~83 TB | — |
172
  | HTML extracted (uncompressed) | 1.9 TB | ~295 TB | — |
173
- | Packed markdown WARC (.md.warc.gz) | ~34.0 GB | ~3.7 TB | **-98.3%** vs HTML |
174
  | Final Parquet (Zstd level 19) | 23.0 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 1.9 TB of uncompressed HTML down to 72.4 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: ~672.8 GB of raw gzipped WARCs becomes **23.0 GB of Parquet** — a **96.6% total reduction** — containing 16,058,095 clean markdown documents.
179
 
180
  ### Processing Times
181
 
182
- Pipeline timings across 830 shards of CC-MAIN-2026-08:
183
 
184
  ```
185
  Download (raw WARC) ████████████████████████ total 48h 2m 49s avg 3m 28s
186
- Convert (HTML → MD) █░░░░░░░░░░░░░░░░░░░░░░░ total 3h 42m 14s avg 16s
187
- Export (Parquet) █████░░░░░░░░░░░░░░░░░░░ total 10h 40m 53s avg 46s
188
- Publish (HuggingFace) ███░░░░░░░░░░░░░░░░░░░░░ total 6h 6m 5s 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 **16,077,531 documents across 831 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 831 files of CC-MAIN-2026-08 (16,077,531 pages total), projected to the full crawl of 100,000 WARC files.
168
 
169
+ | Stage | 831 files (measured) | 100,000 files (projected) | Reduction |
170
  |---|---|---|---|
171
+ | Raw WARC (.warc.gz, downloaded) | ~673.6 GB | ~83 TB | — |
172
  | HTML extracted (uncompressed) | 1.9 TB | ~295 TB | — |
173
+ | Packed markdown WARC (.md.warc.gz) | ~34.1 GB | ~3.7 TB | **-98.3%** vs HTML |
174
  | Final Parquet (Zstd level 19) | 23.0 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 1.9 TB of uncompressed HTML down to 72.5 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: ~673.6 GB of raw gzipped WARCs becomes **23.0 GB of Parquet** — a **96.6% total reduction** — containing 16,077,531 clean markdown documents.
179
 
180
  ### Processing Times
181
 
182
+ Pipeline timings across 831 shards of CC-MAIN-2026-08:
183
 
184
  ```
185
  Download (raw WARC) ████████████████████████ total 48h 2m 49s avg 3m 28s
186
+ Convert (HTML → MD) █░░░░░░░░░░░░░░░░░░░░░░░ total 3h 49m 21s avg 16s
187
+ Export (Parquet) █████░░░░░░░░░░░░░░░░░░░ total 10h 43m 1s avg 46s
188
+ Publish (HuggingFace) ███░░░░░░░░░░░░░░░░░░░░░ total 6h 6m 32s avg 26s
189
  ```
190
 
191
  ### Dataset Charts
data/CC-MAIN-2026-08/00336.parquet ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:073047a12daf5edf523d97f0dce145923540421340943eb195ae10b939da2529
3
+ size 30443713
stats.csv CHANGED
@@ -261,6 +261,7 @@ CC-MAIN-2026-08,332,19576,2569972796,95254532,30402651,2026-03-13T18:10:41Z,0,59
261
  CC-MAIN-2026-08,333,19521,2571725088,93980385,29298182,2026-03-13T18:18:26Z,0,357,104,14
262
  CC-MAIN-2026-08,334,19200,2505425210,91193061,28513905,2026-03-13T18:25:11Z,0,336,73,20
263
  CC-MAIN-2026-08,335,19262,2576133430,95288720,29611938,2026-03-13T18:34:27Z,0,454,88,33
 
264
  CC-MAIN-2026-08,401,19441,2586676962,94334130,29618260,2026-03-13T11:41:00Z,0,0,33,0
265
  CC-MAIN-2026-08,402,19446,2552785517,96654791,30560442,2026-03-13T11:49:21Z,0,443,57,0
266
  CC-MAIN-2026-08,403,19320,2579048144,91406292,29341198,2026-03-13T11:56:46Z,0,414,29,0
@@ -413,7 +414,7 @@ CC-MAIN-2026-08,753,19334,2555127733,89627865,28359904,2026-03-13T18:10:41Z,597,
413
  CC-MAIN-2026-08,754,19329,2566006656,91996350,29169301,2026-03-13T18:18:47Z,382,0,100,28
414
  CC-MAIN-2026-08,755,19381,2559472362,94370252,30420342,2026-03-13T18:25:58Z,330,0,97,28
415
  CC-MAIN-2026-08,756,19494,2576683987,96595694,31300764,2026-03-13T18:35:21Z,412,0,136,28
416
- CC-MAIN-2026-08,757,19539,2604837549,90805230,28789224,2026-03-13T18:43:16Z,369,0,61,0
417
  CC-MAIN-2026-08,851,19575,2612873485,97000364,30911294,2026-03-13T06:41:39Z,163,0,37,204
418
  CC-MAIN-2026-08,852,19334,2546015208,94959847,30205044,2026-03-13T06:45:52Z,183,0,57,11
419
  CC-MAIN-2026-08,853,19694,2567973975,95901932,30303494,2026-03-13T06:50:12Z,150,0,28,80
 
261
  CC-MAIN-2026-08,333,19521,2571725088,93980385,29298182,2026-03-13T18:18:26Z,0,357,104,14
262
  CC-MAIN-2026-08,334,19200,2505425210,91193061,28513905,2026-03-13T18:25:11Z,0,336,73,20
263
  CC-MAIN-2026-08,335,19262,2576133430,95288720,29611938,2026-03-13T18:34:27Z,0,454,88,33
264
+ CC-MAIN-2026-08,336,19436,2580771346,97329551,30443713,2026-03-13T18:44:06Z,0,427,128,0
265
  CC-MAIN-2026-08,401,19441,2586676962,94334130,29618260,2026-03-13T11:41:00Z,0,0,33,0
266
  CC-MAIN-2026-08,402,19446,2552785517,96654791,30560442,2026-03-13T11:49:21Z,0,443,57,0
267
  CC-MAIN-2026-08,403,19320,2579048144,91406292,29341198,2026-03-13T11:56:46Z,0,414,29,0
 
414
  CC-MAIN-2026-08,754,19329,2566006656,91996350,29169301,2026-03-13T18:18:47Z,382,0,100,28
415
  CC-MAIN-2026-08,755,19381,2559472362,94370252,30420342,2026-03-13T18:25:58Z,330,0,97,28
416
  CC-MAIN-2026-08,756,19494,2576683987,96595694,31300764,2026-03-13T18:35:21Z,412,0,136,28
417
+ CC-MAIN-2026-08,757,19539,2604837549,90805230,28789224,2026-03-13T18:43:16Z,369,0,61,27
418
  CC-MAIN-2026-08,851,19575,2612873485,97000364,30911294,2026-03-13T06:41:39Z,163,0,37,204
419
  CC-MAIN-2026-08,852,19334,2546015208,94959847,30205044,2026-03-13T06:45:52Z,183,0,57,11
420
  CC-MAIN-2026-08,853,19694,2567973975,95901932,30303494,2026-03-13T06:50:12Z,150,0,28,80