Commit
·
e99c43f
1
Parent(s):
b6ca68d
Update README.md
Browse files
README.md
CHANGED
|
@@ -30,4 +30,14 @@ model-index:
|
|
| 30 |
metrics:
|
| 31 |
- type: "perplexity" # Required. Example: wer. Use metric id from https://hf.co/metrics
|
| 32 |
value: "46.69" # Required. Example: 20.90
|
| 33 |
-
---
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 30 |
metrics:
|
| 31 |
- type: "perplexity" # Required. Example: wer. Use metric id from https://hf.co/metrics
|
| 32 |
value: "46.69" # Required. Example: 20.90
|
| 33 |
+
---
|
| 34 |
+
|
| 35 |
+
GPT-2 model in luxembourgish language, trained on 636.8 MB of text data, consisting of RTL.lu news articles, comments, parlament speeches, the luxembourgish Wikipedia, Newscrawl, Webcrawl and subtitles.
|
| 36 |
+
The training took place on a 32 GB Nvidia Tesla V100
|
| 37 |
+
with an initial learning rate of 5e-5
|
| 38 |
+
with Batch size 4
|
| 39 |
+
for 109 hours
|
| 40 |
+
for 30 epochs
|
| 41 |
+
|
| 42 |
+
|
| 43 |
+
See the GPT2 model card for considerations on limitations and bias. See the GPT2 documentation for details on GPT2.
|