tamnd commited on
Commit
4323a8e
·
verified ·
1 Parent(s): 4f9b130

Publish shard CC-MAIN-2026-08/00029

Browse files
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 **582,420 documents across 29 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
 
@@ -166,27 +166,27 @@ Empty conversions (pages where trafilatura could not extract meaningful content)
166
 
167
  ### Compression Ratios
168
 
169
- Numbers below are actual measurements summed across all 29 files of CC-MAIN-2026-08 (582,420 pages total), projected to the full crawl of 100,000 WARC files.
170
 
171
- | Stage | 29 files (measured) | 100,000 files (projected) | Reduction |
172
  |---|---|---|---|
173
- | Raw WARC (.warc.gz, downloaded) | ~23.5 GB | ~83 TB | — |
174
- | HTML extracted (uncompressed) | 73.4 GB | ~295 TB | — |
175
- | Packed markdown WARC (.md.warc.gz) | ~1.1 GB | ~3.7 TB | **-98.5%** vs HTML |
176
- | Final Parquet (Zstd level 19) | 796.9 MB | ~2.9 TB | **-30.8%** 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 73.4 GB of uncompressed HTML down to 2.4 GB of markdown — a **98.5% reduction** — before any file-level compression is applied. Parquet with Zstd level 19 then compresses the markdown a further 67.5%.
179
 
180
- End to end: ~23.5 GB of raw gzipped WARCs becomes **796.9 MB of Parquet** — a **96.7% total reduction** — containing 582,420 clean markdown documents.
181
 
182
  ### Processing Times
183
 
184
- Pipeline timings across 29 shards of CC-MAIN-2026-08:
185
 
186
  ```
187
- Pack (download + HTML→MD) ████████████████████████ total 4m 17s avg 8s
188
- Export (Parquet) ██████░░░░░░░░░░░░░░░░░░ total 1m 7s avg 2s
189
- Publish (HuggingFace) ████████████░░░░░░░░░░░░ total 2m 18s avg 4s
190
  ```
191
 
192
  ### Personal and Sensitive Information
 
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 **601,859 documents across 30 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 30 files of CC-MAIN-2026-08 (601,859 pages total), projected to the full crawl of 100,000 WARC files.
170
 
171
+ | Stage | 30 files (measured) | 100,000 files (projected) | Reduction |
172
  |---|---|---|---|
173
+ | Raw WARC (.warc.gz, downloaded) | ~24.3 GB | ~83 TB | — |
174
+ | HTML extracted (uncompressed) | 75.8 GB | ~295 TB | — |
175
+ | Packed markdown WARC (.md.warc.gz) | ~1.2 GB | ~3.7 TB | **-98.5%** vs HTML |
176
+ | Final Parquet (Zstd level 19) | 825.3 MB | ~2.9 TB | **-30.9%** 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 75.8 GB of uncompressed HTML down to 2.5 GB of markdown — a **98.5% reduction** — before any file-level compression is applied. Parquet with Zstd level 19 then compresses the markdown a further 67.5%.
179
 
180
+ End to end: ~24.3 GB of raw gzipped WARCs becomes **825.3 MB of Parquet** — a **96.7% total reduction** — containing 601,859 clean markdown documents.
181
 
182
  ### Processing Times
183
 
184
+ Pipeline timings across 30 shards of CC-MAIN-2026-08:
185
 
186
  ```
187
+ Pack (download + HTML→MD) ████████████████████████ total 5m 29s avg 10s
188
+ Export (Parquet) ██████░░░░░░░░░░░░░░░░░░ total 1m 24s avg 2s
189
+ Publish (HuggingFace) ██████████░░░░░░░░░░░░░░ total 2m 22s avg 4s
190
  ```
191
 
192
  ### Personal and Sensitive Information
charts/compression_pie.png CHANGED
charts/rows_chart.png CHANGED
charts/size_chart.png CHANGED
charts/timing_chart.png CHANGED
data/CC-MAIN-2026-08/00029.parquet ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:9d500c33034992ba811204ac962b47fa726a4fa86a315a09dd0535d90ba3d52d
3
+ size 29839666
stats.csv CHANGED
@@ -27,4 +27,5 @@ CC-MAIN-2026-08,24,19587,2579501303,92876466,30976940,2026-03-13T05:47:58Z,0,0,6
27
  CC-MAIN-2026-08,25,18824,2494782172,91229388,29429097,2026-03-13T05:49:04Z,42,15,4
28
  CC-MAIN-2026-08,26,19453,2545699083,94363861,30370568,2026-03-13T05:50:42Z,73,17,2
29
  CC-MAIN-2026-08,27,19612,2596439798,93518836,30208089,2026-03-13T05:52:18Z,71,18,2
30
- CC-MAIN-2026-08,28,19560,2579397903,97788121,30727702,2026-03-13T05:53:48Z,71,17,0
 
 
27
  CC-MAIN-2026-08,25,18824,2494782172,91229388,29429097,2026-03-13T05:49:04Z,42,15,4
28
  CC-MAIN-2026-08,26,19453,2545699083,94363861,30370568,2026-03-13T05:50:42Z,73,17,2
29
  CC-MAIN-2026-08,27,19612,2596439798,93518836,30208089,2026-03-13T05:52:18Z,71,18,2
30
+ CC-MAIN-2026-08,28,19560,2579397903,97788121,30727702,2026-03-13T05:53:55Z,71,17,4
31
+ CC-MAIN-2026-08,29,19439,2588405263,93704600,29839666,2026-03-13T05:55:26Z,72,17,0