Datasets:
BerenMillidge
commited on
Commit
•
aae9101
1
Parent(s):
6e2848c
Update README.md
Browse files
README.md
CHANGED
@@ -6,22 +6,26 @@ license: odc-by
|
|
6 |
|
7 |
<!-- Provide a quick summary of the dataset. -->
|
8 |
|
9 |
-
Zyda2 is a 5 trillion token language modeling dataset created by collecting open and high quality datasets and combining them and
|
|
|
|
|
10 |
|
11 |
An early version of Zyda2 was used as the primary dataset for phase 1 pretraining of our Zamba2 series [of](Zyphra/Zamba2-2.7B) [models](Zyphra/Zamba2-1.2B) which perform extremely strongly on a per-token basis and are often state-of-the-art for their size, testifying to the strength of Zyda2 as a pretraining dataset.
|
12 |
|
13 |
-
|
|
|
14 |
|
15 |
-
|
16 |
|
|
|
17 |
|
18 |
## How to download
|
19 |
|
20 |
-
// TODO
|
21 |
|
22 |
## Breakdown by component
|
23 |
|
24 |
-
// TODO
|
25 |
|
26 |
### Dataset Description
|
27 |
|
@@ -47,7 +51,7 @@ Dataset fields:
|
|
47 |
|
48 |
### Source Data
|
49 |
|
50 |
-
Zyda2 is comprised of four high quality open-source datasets
|
51 |
|
52 |
Zyda1: https://huggingface.co/datasets/Zyphra/Zyda
|
53 |
|
@@ -60,18 +64,6 @@ FineWeb-Edu-2 https://huggingface.co/datasets/HuggingFaceFW/fineweb-edu
|
|
60 |
|
61 |
// Pie chart of composition -- YURY!
|
62 |
|
63 |
-
#### Data Collection and Processing
|
64 |
-
|
65 |
-
<!-- This section describes the data collection and processing process such as data selection criteria, filtering and normalization methods, tools and libraries used, etc. -->
|
66 |
-
|
67 |
-
Zyda was created using a two stage post-processing pipeline consisting of *filtering* and *deduplication*.
|
68 |
-
|
69 |
-
For the filtering stage, we utilized a set of hand-crafted and tuned filters derived from a number of sources such as C4, RedPajama, and Gopher, in addition to our own filters.
|
70 |
-
|
71 |
-
For the deduplication stage, we used minhash approximate deduplication. We deduplicated on 13-grams and used a minhash signature size of 128 and filtered out documents above a Jaccard similarity of 0.4.
|
72 |
-
|
73 |
-
For full details on our data processing, see the [Zyda technical report](https://arxiv.org/abs/2406.01981) and our [dataset processing code](https://github.com/Zyphra/Zyda_processing).
|
74 |
-
|
75 |
|
76 |
#### Personal and Sensitive Information
|
77 |
|
|
|
6 |
|
7 |
<!-- Provide a quick summary of the dataset. -->
|
8 |
|
9 |
+
Zyda2 is a 5 trillion token language modeling dataset created by collecting open and high quality datasets and combining them and cross-deduplication and model-based quality filtering. Zyda2 comprises diverse sources of web data, highly educational content, math, code, and scientific papers.
|
10 |
+
|
11 |
+
To construct Zyda2, we took the best open-source datasets available: Zyda, FineWeb, DCLM, Dolma. Models trained on Zyda2 significantly outperform identical models trained on the Pile, RefinedWeb, FineWeb, FineWeb-Edu, and DCLM. Due to our post-processing deduplication, filtering, and weighting pipeline, Zyda2 outperforms all its constituent datasets in resulting model quality.
|
12 |
|
13 |
An early version of Zyda2 was used as the primary dataset for phase 1 pretraining of our Zamba2 series [of](Zyphra/Zamba2-2.7B) [models](Zyphra/Zamba2-1.2B) which perform extremely strongly on a per-token basis and are often state-of-the-art for their size, testifying to the strength of Zyda2 as a pretraining dataset.
|
14 |
|
15 |
+
According to our evaluations, Zyda2 is the most performant per-token open dataset available. Zyda2 excels at educational and natural language reasoning content. For code performance, we reccomend mixing it with a pure code dataset such as [Starcoder](https://huggingface.co/bigcode/starcoder).
|
16 |
+
|
17 |
|
18 |
+
// TODO Ablation scores key plots
|
19 |
|
20 |
+
For more information, please see our technical blog (-/TODO LINK)
|
21 |
|
22 |
## How to download
|
23 |
|
24 |
+
// TODO YURY
|
25 |
|
26 |
## Breakdown by component
|
27 |
|
28 |
+
// TODO YURY
|
29 |
|
30 |
### Dataset Description
|
31 |
|
|
|
51 |
|
52 |
### Source Data
|
53 |
|
54 |
+
Zyda2 is comprised of four high quality open-source datasets:
|
55 |
|
56 |
Zyda1: https://huggingface.co/datasets/Zyphra/Zyda
|
57 |
|
|
|
64 |
|
65 |
// Pie chart of composition -- YURY!
|
66 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
67 |
|
68 |
#### Personal and Sensitive Information
|
69 |
|