Update README.md
Browse files
README.md
CHANGED
|
@@ -20,6 +20,7 @@ language:
|
|
| 20 |
- da
|
| 21 |
size_categories:
|
| 22 |
- 100M<n<1B
|
|
|
|
| 23 |
---
|
| 24 |
|
| 25 |
# Details
|
|
@@ -63,6 +64,10 @@ Note how the number of permutations is lower, a lower batch size and we have low
|
|
| 63 |
We encountered several OOM errors that were a bit inexplicable and decided to lower the memory footprint in this way.
|
| 64 |
The hardware we used was a machine with 128 cores and 1TB of RAM. This data should take less than 100GB of disk space.
|
| 65 |
|
|
|
|
|
|
|
|
|
|
|
|
|
| 66 |
|
| 67 |
## Citation
|
| 68 |
|
|
|
|
| 20 |
- da
|
| 21 |
size_categories:
|
| 22 |
- 100M<n<1B
|
| 23 |
+
license: unknown
|
| 24 |
---
|
| 25 |
|
| 26 |
# Details
|
|
|
|
| 64 |
We encountered several OOM errors that were a bit inexplicable and decided to lower the memory footprint in this way.
|
| 65 |
The hardware we used was a machine with 128 cores and 1TB of RAM. This data should take less than 100GB of disk space.
|
| 66 |
|
| 67 |
+
## Licenses
|
| 68 |
+
|
| 69 |
+
For licensing of the data, we refer to the paper. Specifically for the Twitter data, we made an assumption that the data has been extracted using a version of the Twitter API which is usually MIT-licensed, to consider it as "an upper-bound".
|
| 70 |
+
Though, we understand that the software/code is usually MIT-licensed. Feel free to leave the Twitter data out. Other Twitter datasets here on HF seem to have a flavor of cc-by*.
|
| 71 |
|
| 72 |
## Citation
|
| 73 |
|