Publish shard CC-MAIN-2026-08/00371
Browse files- README.md +10 -10
- data/CC-MAIN-2026-08/00371.parquet +3 -0
- stats.csv +1 -0
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 **21,
|
| 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 |
|
|
@@ -164,27 +164,27 @@ 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) | 2.6 TB | ~295 TB | — |
|
| 173 |
| Packed markdown WARC (.md.warc.gz) | ~45.8 GB | ~3.7 TB | **-98.3%** vs HTML |
|
| 174 |
-
| Final Parquet (Zstd level 19) |
|
| 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 2.6 TB of uncompressed HTML down to 97.
|
| 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 5h 55m 4s avg 19s
|
| 187 |
-
Export (Parquet) █████░░░░░░░░░░░░░░░░░░░ total 20h
|
| 188 |
Publish (HuggingFace) ██░░░░░░░░░░░░░░░░░░░░░░ total 7h 18m 25s avg 23s
|
| 189 |
```
|
| 190 |
|
|
|
|
| 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 **21,626,337 documents across 1119 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 1119 files of CC-MAIN-2026-08 (21,626,337 pages total), projected to the full crawl of 100,000 WARC files.
|
| 168 |
|
| 169 |
+
| Stage | 1119 files (measured) | 100,000 files (projected) | Reduction |
|
| 170 |
|---|---|---|---|
|
| 171 |
+
| Raw WARC (.warc.gz, downloaded) | ~907.0 GB | ~83 TB | — |
|
| 172 |
| HTML extracted (uncompressed) | 2.6 TB | ~295 TB | — |
|
| 173 |
| Packed markdown WARC (.md.warc.gz) | ~45.8 GB | ~3.7 TB | **-98.3%** vs HTML |
|
| 174 |
+
| Final Parquet (Zstd level 19) | 31.0 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 2.6 TB of uncompressed HTML down to 97.6 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: ~907.0 GB of raw gzipped WARCs becomes **31.0 GB of Parquet** — a **96.6% total reduction** — containing 21,626,337 clean markdown documents.
|
| 179 |
|
| 180 |
### Processing Times
|
| 181 |
|
| 182 |
+
Pipeline timings across 1119 shards of CC-MAIN-2026-08:
|
| 183 |
|
| 184 |
```
|
| 185 |
+
Download (raw WARC) ████████████████████████ total 85h 5m 8s avg 4m 33s
|
| 186 |
Convert (HTML → MD) █░░░░░░░░░░░░░░░░░░░░░░░ total 5h 55m 4s avg 19s
|
| 187 |
+
Export (Parquet) █████░░░░░░░░░░░░░░░░░░░ total 20h 32m 19s avg 1m 6s
|
| 188 |
Publish (HuggingFace) ██░░░░░░░░░░░░░░░░░░░░░░ total 7h 18m 25s avg 23s
|
| 189 |
```
|
| 190 |
|
data/CC-MAIN-2026-08/00371.parquet
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:712932c9ec9fbd8800786fd5425e09573b8375ec4e8b22c4631e074033153195
|
| 3 |
+
size 29867038
|
stats.csv
CHANGED
|
@@ -332,6 +332,7 @@ CC-MAIN-2026-08,367,19224,2546537039,93497264,29177902,2026-03-14T01:06:44Z,437,
|
|
| 332 |
CC-MAIN-2026-08,368,19258,2574726313,93452151,30093839,2026-03-14T01:06:44Z,756,0,92,5
|
| 333 |
CC-MAIN-2026-08,369,19299,2562815931,93550342,29772834,2026-03-14T01:06:44Z,777,0,141,5
|
| 334 |
CC-MAIN-2026-08,370,19242,2559224128,97323871,31587291,2026-03-14T01:14:04Z,750,0,59,23
|
|
|
|
| 335 |
CC-MAIN-2026-08,401,19441,2586676962,94334130,29618260,2026-03-13T11:41:00Z,0,0,33,0
|
| 336 |
CC-MAIN-2026-08,402,19446,2552785517,96654791,30560442,2026-03-13T11:49:21Z,0,443,57,0
|
| 337 |
CC-MAIN-2026-08,403,19320,2579048144,91406292,29341198,2026-03-13T11:56:46Z,0,414,29,0
|
|
|
|
| 332 |
CC-MAIN-2026-08,368,19258,2574726313,93452151,30093839,2026-03-14T01:06:44Z,756,0,92,5
|
| 333 |
CC-MAIN-2026-08,369,19299,2562815931,93550342,29772834,2026-03-14T01:06:44Z,777,0,141,5
|
| 334 |
CC-MAIN-2026-08,370,19242,2559224128,97323871,31587291,2026-03-14T01:14:04Z,750,0,59,23
|
| 335 |
+
CC-MAIN-2026-08,371,19658,2574470616,95555396,29867038,2026-03-14T01:22:05Z,492,0,26,0
|
| 336 |
CC-MAIN-2026-08,401,19441,2586676962,94334130,29618260,2026-03-13T11:41:00Z,0,0,33,0
|
| 337 |
CC-MAIN-2026-08,402,19446,2552785517,96654791,30560442,2026-03-13T11:49:21Z,0,443,57,0
|
| 338 |
CC-MAIN-2026-08,403,19320,2579048144,91406292,29341198,2026-03-13T11:56:46Z,0,414,29,0
|