BerenMillidge
commited on
Update README.md
Browse files
README.md
CHANGED
@@ -63,11 +63,7 @@ configs:
|
|
63 |
|
64 |
<!-- Provide a quick summary of the dataset. -->
|
65 |
|
66 |
-
|
67 |
-
|
68 |
-
## Dataset Details
|
69 |
-
|
70 |
-
Dataset is created by filtering and deduplicating openly available datasets.
|
71 |
|
72 |
|
73 |
## How to download
|
@@ -93,34 +89,9 @@ For downloading individual components put their name in the name arg of `load_da
|
|
93 |
<!-- Provide a longer summary of what this dataset is. -->
|
94 |
|
95 |
- **Curated by:** Zyphra
|
96 |
-
- **
|
97 |
-
- **Shared by [optional]:** [More Information Needed]
|
98 |
-
- **Language(s) (NLP):** English
|
99 |
- **License:** Open Data Commons License
|
100 |
|
101 |
-
### Dataset Sources [optional]
|
102 |
-
|
103 |
-
<!-- Provide the basic links for the dataset. -->
|
104 |
-
|
105 |
-
- **Repository:** [More Information Needed]
|
106 |
-
- **Paper [optional]:** [More Information Needed]
|
107 |
-
- **Demo [optional]:** [More Information Needed]
|
108 |
-
|
109 |
-
## Uses
|
110 |
-
|
111 |
-
<!-- Address questions around how the dataset is intended to be used. -->
|
112 |
-
|
113 |
-
### Direct Use
|
114 |
-
|
115 |
-
<!-- This section describes suitable use cases for the dataset. -->
|
116 |
-
|
117 |
-
[More Information Needed]
|
118 |
-
|
119 |
-
### Out-of-Scope Use
|
120 |
-
|
121 |
-
<!-- This section addresses misuse, malicious use, and uses that the dataset will not work well for. -->
|
122 |
-
|
123 |
-
[More Information Needed]
|
124 |
|
125 |
## Dataset Structure
|
126 |
|
@@ -133,19 +104,9 @@ Dataset fields:
|
|
133 |
- `source_other`: metadata from the source dataset (converted to json string)
|
134 |
|
135 |
|
136 |
-
[More Information Needed]
|
137 |
-
|
138 |
-
## Dataset Creation
|
139 |
-
|
140 |
-
### Curation Rationale
|
141 |
-
|
142 |
-
<!-- Motivation for the creation of this dataset. -->
|
143 |
-
|
144 |
-
[More Information Needed]
|
145 |
|
146 |
### Source Data
|
147 |
|
148 |
-
<!-- This section describes the source data (e.g. news text and headlines, social media posts, translated sentences, ...). -->
|
149 |
|
150 |
Pile Uncopyrighted: https://huggingface.co/datasets/monology/pile-uncopyrighted
|
151 |
|
@@ -167,72 +128,19 @@ StarCoder: https://huggingface.co/datasets/bigcode/starcoderdata
|
|
167 |
|
168 |
[More Information Needed]
|
169 |
|
170 |
-
#### Who are the source data producers?
|
171 |
-
|
172 |
-
<!-- This section describes the people or systems who originally created the data. It should also include self-reported demographic or identity information for the source data creators if this information is available. -->
|
173 |
-
|
174 |
-
[More Information Needed]
|
175 |
-
|
176 |
-
### Annotations [optional]
|
177 |
-
|
178 |
-
<!-- If the dataset contains annotations which are not part of the initial data collection, use this section to describe them. -->
|
179 |
|
180 |
-
#### Annotation process
|
181 |
-
|
182 |
-
<!-- This section describes the annotation process such as annotation tools used in the process, the amount of data annotated, annotation guidelines provided to the annotators, interannotator statistics, annotation validation, etc. -->
|
183 |
-
|
184 |
-
[More Information Needed]
|
185 |
|
186 |
-
#### Who are the annotators?
|
187 |
-
|
188 |
-
<!-- This section describes the people or systems who created the annotations. -->
|
189 |
-
|
190 |
-
[More Information Needed]
|
191 |
|
192 |
#### Personal and Sensitive Information
|
193 |
|
194 |
-
|
195 |
-
|
196 |
-
[More Information Needed]
|
197 |
|
198 |
## Bias, Risks, and Limitations
|
199 |
|
200 |
-
|
201 |
-
|
202 |
-
[More Information Needed]
|
203 |
-
|
204 |
-
### Recommendations
|
205 |
-
|
206 |
-
<!-- This section is meant to convey recommendations with respect to the bias, risk, and technical limitations. -->
|
207 |
-
|
208 |
-
Users should be made aware of the risks, biases and limitations of the dataset. More information needed for further recommendations.
|
209 |
|
210 |
## Citation [optional]
|
211 |
|
212 |
-
|
213 |
-
|
214 |
-
**BibTeX:**
|
215 |
-
|
216 |
-
[More Information Needed]
|
217 |
-
|
218 |
-
**APA:**
|
219 |
-
|
220 |
-
[More Information Needed]
|
221 |
-
|
222 |
-
## Glossary [optional]
|
223 |
-
|
224 |
-
<!-- If relevant, include terms and calculations in this section that can help readers understand the dataset or dataset card. -->
|
225 |
-
|
226 |
-
[More Information Needed]
|
227 |
-
|
228 |
-
## More Information [optional]
|
229 |
-
|
230 |
-
[More Information Needed]
|
231 |
-
|
232 |
-
## Dataset Card Authors [optional]
|
233 |
-
|
234 |
-
[More Information Needed]
|
235 |
-
|
236 |
-
## Dataset Card Contact
|
237 |
|
238 |
-
|
|
|
63 |
|
64 |
<!-- Provide a quick summary of the dataset. -->
|
65 |
|
66 |
+
Zyda is a 1.3T language modelling dataset created by collecting open and high quality datasets and combining them and performing a uniform filtering and deduplication step. We find that Zyda performs extremely well in ablations and is at least comparable and potentially better to the best openly available datasets available, due to our meticulous post-processing pipeline. We think the best use of Zyda is either as a standalone dataset for language model training up to the 1T scale, or in combination with Fineweb or Dolma for multi-trillion token training.
|
|
|
|
|
|
|
|
|
67 |
|
68 |
|
69 |
## How to download
|
|
|
89 |
<!-- Provide a longer summary of what this dataset is. -->
|
90 |
|
91 |
- **Curated by:** Zyphra
|
92 |
+
- **Language(s) (NLP):** Primarily English
|
|
|
|
|
93 |
- **License:** Open Data Commons License
|
94 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
95 |
|
96 |
## Dataset Structure
|
97 |
|
|
|
104 |
- `source_other`: metadata from the source dataset (converted to json string)
|
105 |
|
106 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
107 |
|
108 |
### Source Data
|
109 |
|
|
|
110 |
|
111 |
Pile Uncopyrighted: https://huggingface.co/datasets/monology/pile-uncopyrighted
|
112 |
|
|
|
128 |
|
129 |
[More Information Needed]
|
130 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
131 |
|
|
|
|
|
|
|
|
|
|
|
132 |
|
|
|
|
|
|
|
|
|
|
|
133 |
|
134 |
#### Personal and Sensitive Information
|
135 |
|
136 |
+
As a language modelling dataset, it likely contains PII which has not been filtered out of the component datasets and which may have been missed by our own filters.
|
|
|
|
|
137 |
|
138 |
## Bias, Risks, and Limitations
|
139 |
|
140 |
+
As a dataset comprised of open web scrapes, it is likely that it contains biased and toxic content.
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
141 |
|
142 |
## Citation [optional]
|
143 |
|
144 |
+
If you use our dataset to train a model, please cite us at:
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
145 |
|
146 |
+
(-/TODO)
|