Commit
•
da11db9
1
Parent(s):
6ae7558
Update README.md
Browse files
README.md
CHANGED
|
@@ -71,7 +71,7 @@ This loads the model at checkpoint 500.
|
|
| 71 |
|
| 72 |
### Training Data
|
| 73 |
|
| 74 |
-
This model was trained on the 100M-word BabyLM dataset.
|
| 75 |
Before training, we first transform the dataset into
|
| 76 |
the corresponding impossible language, as described in
|
| 77 |
our paper.
|
|
@@ -88,7 +88,6 @@ rate that linearly warms up from 0 to 6e-4 over 300 steps.
|
|
| 88 |
- **Hours used:** ~24 hours.
|
| 89 |
|
| 90 |
## Citation
|
| 91 |
-
**BibTeX:**
|
| 92 |
|
| 93 |
```bibtex
|
| 94 |
@inproceedings{kallini-etal-2024-mission,
|
|
@@ -112,8 +111,6 @@ rate that linearly warms up from 0 to 6e-4 over 300 steps.
|
|
| 112 |
}
|
| 113 |
```
|
| 114 |
|
| 115 |
-
**APA:**
|
| 116 |
-
|
| 117 |
## Model Card Authors
|
| 118 |
|
| 119 |
Julie Kallini
|
|
|
|
| 71 |
|
| 72 |
### Training Data
|
| 73 |
|
| 74 |
+
This model was trained on the [100M-word BabyLM dataset](https://babylm.github.io/).
|
| 75 |
Before training, we first transform the dataset into
|
| 76 |
the corresponding impossible language, as described in
|
| 77 |
our paper.
|
|
|
|
| 88 |
- **Hours used:** ~24 hours.
|
| 89 |
|
| 90 |
## Citation
|
|
|
|
| 91 |
|
| 92 |
```bibtex
|
| 93 |
@inproceedings{kallini-etal-2024-mission,
|
|
|
|
| 111 |
}
|
| 112 |
```
|
| 113 |
|
|
|
|
|
|
|
| 114 |
## Model Card Authors
|
| 115 |
|
| 116 |
Julie Kallini
|