Clarify Model Description and Add Project Page Link
#2
by
nielsr
HF Staff
- opened
README.md
CHANGED
@@ -1,20 +1,20 @@
|
|
1 |
---
|
2 |
-
license: apache-2.0
|
3 |
datasets:
|
4 |
- togethercomputer/RedPajama-Data-1T
|
5 |
language:
|
6 |
- en
|
7 |
-
pipeline_tag: text-generation
|
8 |
library_name: transformers
|
|
|
|
|
9 |
---
|
10 |
|
11 |
## PDS-470M
|
12 |
|
13 |
-
[paper](https://arxiv.org/abs/2410.07064) | [code](https://github.com/microsoft/LMOps/tree/main/data_selection)
|
14 |
|
15 |
-
**PDS-470M** is a 470M
|
16 |
|
17 |
-
The PDS framework is based on the [Pontryagin's maximum principle](https://en.wikipedia.org/wiki/Pontryagin%27s_maximum_principle#:~:text=Pontryagin's%20maximum%20principle%20is%20used,the%20state%20or%20input%20controls.) for optimal pre-training data selection,
|
18 |
|
19 |
Please refer to our [paper](https://arxiv.org/abs/2410.07064) for more details.
|
20 |
|
@@ -51,4 +51,4 @@ PDS-selected data improves the performance of language models pre-trained from s
|
|
51 |
journal={arXiv preprint arXiv:2410.07064},
|
52 |
year={2024}
|
53 |
}
|
54 |
-
```
|
|
|
1 |
---
|
|
|
2 |
datasets:
|
3 |
- togethercomputer/RedPajama-Data-1T
|
4 |
language:
|
5 |
- en
|
|
|
6 |
library_name: transformers
|
7 |
+
license: apache-2.0
|
8 |
+
pipeline_tag: text-generation
|
9 |
---
|
10 |
|
11 |
## PDS-470M
|
12 |
|
13 |
+
[paper](https://arxiv.org/abs/2410.07064) | [code](https://github.com/microsoft/LMOps/tree/main/data_selection) | [project page](https://github.com/microsoft/LMOps/tree/main/data_selection)
|
14 |
|
15 |
+
**PDS-470M** is a 470M parameter Mistral architecture model **pretrained from scratch** using the PDS framework on data selected from the CC split of [Redpajama](https://github.com/togethercomputer/RedPajama-Data).
|
16 |
|
17 |
+
The PDS framework is based on the [Pontryagin's maximum principle](https://en.wikipedia.org/wiki/Pontryagin%27s_maximum_principle#:~:text=Pontryagin's%20maximum%20principle%20is%20used,the%20state%20or%20input%20controls.) for optimal pre-training data selection, offering strong theoretical support and scalability for training large language models.
|
18 |
|
19 |
Please refer to our [paper](https://arxiv.org/abs/2410.07064) for more details.
|
20 |
|
|
|
51 |
journal={arXiv preprint arXiv:2410.07064},
|
52 |
year={2024}
|
53 |
}
|
54 |
+
```
|