Datasets:

Modalities:
Text
ArXiv:
Tags:
Maurice Weber commited on
Commit
bd77c17
1 Parent(s): 1a99058

add raw doc+token counts

Browse files
Files changed (1) hide show
  1. README.md +26 -14
README.md CHANGED
@@ -15,7 +15,7 @@ pretty_name: Red Pajama V2 Dataset
15
  RedPajama-V2 is an open dataset for training large language models. The dataset includes over 100B text
16
  documents coming from 84 CommonCrawl snapshots and processed using
17
  the [CCNet](https://github.com/facebookresearch/cc_net) pipeline. Out of these, there are 30B documents in the corpus
18
- that additionally come with quality signals. In addition, we also provide the ids of duplicated documents which can be
19
  used to create a dataset with 20B deduplicated documents.
20
 
21
  Check out our [blog post](https://together.ai/blog/redpajama-data-v2) for more details on the build process, dataset
@@ -30,8 +30,9 @@ ds = load_dataset("togethercomputer/RedPajama-Data-V2", name="sample")
30
  ```
31
 
32
  To download a the dataset for a specific combination of `{partition} x {snapshot_id} x {language}` (e.g., English and
33
- German data from the `head_middle` partition of the 2023-06 and the 2022-49 dumps), you can run the following command.
34
- _Note that this will downlaod the entire dumps and requires ~1TB disk space per dump_.
 
35
 
36
  ```python
37
  from datasets import load_dataset
@@ -71,7 +72,7 @@ done <"$listings_file"
71
 
72
  ```
73
 
74
- In addition, for the `head_middle` partition, you can also download the quality signals, minhash signatures and
75
  duplicate ids using the following commands:
76
 
77
  ```bash
@@ -227,16 +228,27 @@ RedPajama-V2 is an open dataset for training large language models and includes
227
  | minhash_signature_0.9 | Banded minhash signature of the document, for fuzzy deduplication at Jaccard similarity 0.9. The signature is based on 128 hash functions and grouped into 5 bands and 25 rows for LSH.. | Deduplication |
228
  | minhash_signature_1.0 | Banded minhash signature of the document, for fuzzy deduplication at Jaccard similarity 1.0. The signature is based on 128 hash functions and grouped into 1 band and 128 rows for LSH. | Deduplication |
229
 
230
- #### Document and Token Counts for the Annotated and deduplicated `head_middle` part of the dataset
231
-
232
- | | # Documents | Estimated Token count (deduped) |
233
- |-------|-------------|---------------------------------|
234
- | en | 14.5B | 20.5T |
235
- | de | 1.9B | 3.0T |
236
- | fr | 1.6B | 2.7T |
237
- | es | 1.8B | 2.8T |
238
- | it | 0.9B | 1.5T |
239
- | Total | 20.8B | 30.4T |
 
 
 
 
 
 
 
 
 
 
 
240
 
241
  ### Languages
242
 
 
15
  RedPajama-V2 is an open dataset for training large language models. The dataset includes over 100B text
16
  documents coming from 84 CommonCrawl snapshots and processed using
17
  the [CCNet](https://github.com/facebookresearch/cc_net) pipeline. Out of these, there are 30B documents in the corpus
18
+ that additionally come with quality signals. In addition, we also provide the ids of duplicated documents which can be
19
  used to create a dataset with 20B deduplicated documents.
20
 
21
  Check out our [blog post](https://together.ai/blog/redpajama-data-v2) for more details on the build process, dataset
 
30
  ```
31
 
32
  To download a the dataset for a specific combination of `{partition} x {snapshot_id} x {language}` (e.g., English and
33
+ German data from the `head_middle` partition of the 2023-06 and the 2022-49 dumps), you can run the following command
34
+ which downloads the raw (i.e., not deduplicated) part of the dataset.
35
+ _Note that this will download the entire dumps and requires ~1TB disk space per dump_.
36
 
37
  ```python
38
  from datasets import load_dataset
 
72
 
73
  ```
74
 
75
+ In addition, for the `head_middle` partition, you can also download the quality signals, minhash signatures and
76
  duplicate ids using the following commands:
77
 
78
  ```bash
 
228
  | minhash_signature_0.9 | Banded minhash signature of the document, for fuzzy deduplication at Jaccard similarity 0.9. The signature is based on 128 hash functions and grouped into 5 bands and 25 rows for LSH.. | Deduplication |
229
  | minhash_signature_1.0 | Banded minhash signature of the document, for fuzzy deduplication at Jaccard similarity 1.0. The signature is based on 128 hash functions and grouped into 1 band and 128 rows for LSH. | Deduplication |
230
 
231
+ #### Raw Document and Token Counts (`head_middle`)
232
+
233
+ | | # Documents (deduped) | Estimated Token count (deduped) |
234
+ |-------|-----------------------|---------------------------------|
235
+ | en | 24.5B | 37.0T |
236
+ | de | 2.7B | 4.1T |
237
+ | fr | 2.2B | 3.7T |
238
+ | es | 2.3B | 3.9T |
239
+ | it | 1.2B | 1.9T |
240
+ | Total | 32.9B | 50.6T |
241
+
242
+ #### Deduplicated Document and Token Counts (`head_middle`)
243
+
244
+ | | # Documents (total) | Estimated Token count (total) |
245
+ |-------|---------------------|-------------------------------|
246
+ | en | 14.5B | 20.5T |
247
+ | de | 1.9B | 3.0T |
248
+ | fr | 1.6B | 2.7T |
249
+ | es | 1.8B | 2.8T |
250
+ | it | 0.9B | 1.5T |
251
+ | Total | 20.8B | 30.4T |
252
 
253
  ### Languages
254