NOKUBI Takatsugu
commited on
Commit
•
97a67a5
1
Parent(s):
9f0afec
license: mit
Browse files
README.md
CHANGED
|
@@ -33,6 +33,10 @@ Using a2-highgpu-4 instance (A100 x4), it takes about 4 months with some stoppin
|
|
| 33 |
The model gets about 40 perplexity with Wikipedia corpus.
|
| 34 |
The teacher model rinna/japanese-gpt2-meduim gets about 27 perplexity, so the student model is worse.
|
| 35 |
|
|
|
|
|
|
|
|
|
|
|
|
|
| 36 |
---
|
| 37 |
-
license: mit
|
| 38 |
---
|
|
|
|
| 33 |
The model gets about 40 perplexity with Wikipedia corpus.
|
| 34 |
The teacher model rinna/japanese-gpt2-meduim gets about 27 perplexity, so the student model is worse.
|
| 35 |
|
| 36 |
+
# LICENSE
|
| 37 |
+
|
| 38 |
+
MIT (same as rinna/japanese-gpt2-medium)
|
| 39 |
+
|
| 40 |
---
|
| 41 |
+
license: mit
|
| 42 |
---
|