Publish shard CC-MAIN-2026-08/01933
Browse files- README.md +10 -10
- data/CC-MAIN-2026-08/01933.parquet +3 -0
- 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 **14,
|
| 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
|
| 168 |
|
| 169 |
-
| Stage |
|
| 170 |
|---|---|---|---|
|
| 171 |
-
| Raw WARC (.warc.gz, downloaded) | ~
|
| 172 |
| HTML extracted (uncompressed) | 1.7 TB | ~295 TB | — |
|
| 173 |
| Packed markdown WARC (.md.warc.gz) | ~30.0 GB | ~3.7 TB | **-98.3%** vs HTML |
|
| 174 |
| Final Parquet (Zstd level 19) | 20.3 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 1.7 TB of uncompressed HTML down to 63.
|
| 177 |
|
| 178 |
-
End to end: ~
|
| 179 |
|
| 180 |
### Processing Times
|
| 181 |
|
| 182 |
-
Pipeline timings across
|
| 183 |
|
| 184 |
```
|
| 185 |
-
Download (raw WARC) ████████████████████████ total
|
| 186 |
Convert (HTML → MD) █░░░░░░░░░░░░░░░░░░░░░░░ total 1h 50m 33s avg 9s
|
| 187 |
-
Export (Parquet) █████░░░░░░░░░░░░░░░░░░░ total 8h
|
| 188 |
-
Publish (HuggingFace) ███░░░░░░░░░░░░░░░░░░░░░ total 5h 30m
|
| 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 **14,169,602 documents across 733 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 733 files of CC-MAIN-2026-08 (14,169,602 pages total), projected to the full crawl of 100,000 WARC files.
|
| 168 |
|
| 169 |
+
| Stage | 733 files (measured) | 100,000 files (projected) | Reduction |
|
| 170 |
|---|---|---|---|
|
| 171 |
+
| Raw WARC (.warc.gz, downloaded) | ~594.1 GB | ~83 TB | — |
|
| 172 |
| HTML extracted (uncompressed) | 1.7 TB | ~295 TB | — |
|
| 173 |
| Packed markdown WARC (.md.warc.gz) | ~30.0 GB | ~3.7 TB | **-98.3%** vs HTML |
|
| 174 |
| Final Parquet (Zstd level 19) | 20.3 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 1.7 TB of uncompressed HTML down to 63.9 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: ~594.1 GB of raw gzipped WARCs becomes **20.3 GB of Parquet** — a **96.6% total reduction** — containing 14,169,602 clean markdown documents.
|
| 179 |
|
| 180 |
### Processing Times
|
| 181 |
|
| 182 |
+
Pipeline timings across 733 shards of CC-MAIN-2026-08:
|
| 183 |
|
| 184 |
```
|
| 185 |
+
Download (raw WARC) ████████████████████████ total 40h 8m 28s avg 3m 17s
|
| 186 |
Convert (HTML → MD) █░░░░░░░░░░░░░░░░░░░░░░░ total 1h 50m 33s avg 9s
|
| 187 |
+
Export (Parquet) █████░░░░░░░░░░░░░░░░░░░ total 8h 49m 17s avg 43s
|
| 188 |
+
Publish (HuggingFace) ███░░░░░░░░░░░░░░░░░░░░░ total 5h 30m 38s avg 27s
|
| 189 |
```
|
| 190 |
|
| 191 |
### Dataset Charts
|
data/CC-MAIN-2026-08/01933.parquet
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:5cfb6324d67f54a36f6719caa8a9870150939d3e10c6d0af778044ec656115bb
|
| 3 |
+
size 31093153
|
stats.csv
CHANGED
|
@@ -613,6 +613,7 @@ CC-MAIN-2026-08,1929,19636,2583727825,94857326,29780590,2026-03-13T15:21:56Z,463
|
|
| 613 |
CC-MAIN-2026-08,1930,19031,2504624971,92471516,28878497,2026-03-13T15:26:00Z,471,0,45,14
|
| 614 |
CC-MAIN-2026-08,1931,19494,2557608260,100036397,31331643,2026-03-13T15:37:22Z,625,0,55,13
|
| 615 |
CC-MAIN-2026-08,1932,19369,2564301103,93801517,29242281,2026-03-13T15:46:56Z,546,0,30,24
|
|
|
|
| 616 |
CC-MAIN-2026-08,2101,19421,2568457366,93149947,28748313,2026-03-13T13:54:07Z,262,0,31,0
|
| 617 |
CC-MAIN-2026-08,2102,19717,2608848648,96373346,30639542,2026-03-13T14:00:38Z,281,0,109,0
|
| 618 |
CC-MAIN-2026-08,2103,19325,2580718688,94798987,30249683,2026-03-13T14:06:18Z,304,0,34,0
|
|
@@ -639,7 +640,7 @@ CC-MAIN-2026-08,2123,19351,2583952444,90760742,29025397,2026-03-13T15:26:19Z,472
|
|
| 639 |
CC-MAIN-2026-08,2124,19507,2598213424,95309823,30481471,2026-03-13T15:37:22Z,577,0,83,13
|
| 640 |
CC-MAIN-2026-08,2125,19427,2585686106,95664309,29507764,2026-03-13T15:46:11Z,470,0,65,14
|
| 641 |
CC-MAIN-2026-08,2126,19604,2590610069,97476800,30121196,2026-03-13T15:50:56Z,210,0,66,25
|
| 642 |
-
CC-MAIN-2026-08,2127,19399,2549889958,95568982,29906356,2026-03-13T15:58:26Z,317,0,127,
|
| 643 |
CC-MAIN-2026-08,2301,19498,2589050373,94698511,30163652,2026-03-13T13:54:08Z,261,0,33,0
|
| 644 |
CC-MAIN-2026-08,2302,19595,2575185844,96075881,29961924,2026-03-13T14:00:40Z,280,0,112,0
|
| 645 |
CC-MAIN-2026-08,2303,19594,2612324314,94525336,29944034,2026-03-13T14:06:16Z,305,0,30,0
|
|
|
|
| 613 |
CC-MAIN-2026-08,1930,19031,2504624971,92471516,28878497,2026-03-13T15:26:00Z,471,0,45,14
|
| 614 |
CC-MAIN-2026-08,1931,19494,2557608260,100036397,31331643,2026-03-13T15:37:22Z,625,0,55,13
|
| 615 |
CC-MAIN-2026-08,1932,19369,2564301103,93801517,29242281,2026-03-13T15:46:56Z,546,0,30,24
|
| 616 |
+
CC-MAIN-2026-08,1933,19248,2579893230,98704471,31093153,2026-03-13T15:59:11Z,598,0,126,0
|
| 617 |
CC-MAIN-2026-08,2101,19421,2568457366,93149947,28748313,2026-03-13T13:54:07Z,262,0,31,0
|
| 618 |
CC-MAIN-2026-08,2102,19717,2608848648,96373346,30639542,2026-03-13T14:00:38Z,281,0,109,0
|
| 619 |
CC-MAIN-2026-08,2103,19325,2580718688,94798987,30249683,2026-03-13T14:06:18Z,304,0,34,0
|
|
|
|
| 640 |
CC-MAIN-2026-08,2124,19507,2598213424,95309823,30481471,2026-03-13T15:37:22Z,577,0,83,13
|
| 641 |
CC-MAIN-2026-08,2125,19427,2585686106,95664309,29507764,2026-03-13T15:46:11Z,470,0,65,14
|
| 642 |
CC-MAIN-2026-08,2126,19604,2590610069,97476800,30121196,2026-03-13T15:50:56Z,210,0,66,25
|
| 643 |
+
CC-MAIN-2026-08,2127,19399,2549889958,95568982,29906356,2026-03-13T15:58:26Z,317,0,127,13
|
| 644 |
CC-MAIN-2026-08,2301,19498,2589050373,94698511,30163652,2026-03-13T13:54:08Z,261,0,33,0
|
| 645 |
CC-MAIN-2026-08,2302,19595,2575185844,96075881,29961924,2026-03-13T14:00:40Z,280,0,112,0
|
| 646 |
CC-MAIN-2026-08,2303,19594,2612324314,94525336,29944034,2026-03-13T14:06:16Z,305,0,30,0
|