Don Mahurin
commited on
Commit
•
9c30a5b
1
Parent(s):
0543a04
update README
Browse files
README.md
CHANGED
@@ -1,3 +1,34 @@
|
|
1 |
---
|
2 |
license: mit
|
3 |
---
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
---
|
2 |
license: mit
|
3 |
---
|
4 |
+
|
5 |
+
# Language model trained with reversed words
|
6 |
+
|
7 |
+
The model weights here were trained using Tiny Stories, but with the words and puncuation reversed.
|
8 |
+
|
9 |
+
This is described at:
|
10 |
+
|
11 |
+
https://openright.org/reversed-llm-training/
|
12 |
+
|
13 |
+
## Running the reverse inference
|
14 |
+
|
15 |
+
The model may be ran using these weights with the 'llama2.c' project.
|
16 |
+
|
17 |
+
The input tokens are expected in reverse, and the output tokens are also in reverse. For example, "This is a test.", becomes ".test a is This".
|
18 |
+
|
19 |
+
We can simply run with reversed input, and see the reversed output.
|
20 |
+
|
21 |
+
```
|
22 |
+
./run revstories15M.bin -s 3 -i '.sky the in butterfly'
|
23 |
+
```
|
24 |
+
|
25 |
+
> .sky the in butterfly purple beautiful the about family her tell
|
26 |
+
> ...
|
27 |
+
|
28 |
+
But to have the input and output unreversed, we can use the 'wtac' script to reverse the words.
|
29 |
+
|
30 |
+
```
|
31 |
+
./run revstories15M.bin -s 3 -i "$(echo 'butterfly in the sky.' | python3 wtac.py)" | python3 ./wtac.py
|
32 |
+
```
|
33 |
+
|
34 |
+
> Once upon a time, there was a little girl named Lily. She loved to play in the park with her friends. One day, the butterfly landed on Lily's hands and led her to a flower. Lily was very happy and couldn't wait to tell her family about the beautiful purple butterfly in the sky.
|
wtac.py
CHANGED
@@ -1,4 +1,4 @@
|
|
1 |
-
#!/usr/bin/python3
|
2 |
# reverses word and punctuation order.
|
3 |
import re
|
4 |
import sys
|
|
|
1 |
+
#!/usr/bin/env python3
|
2 |
# reverses word and punctuation order.
|
3 |
import re
|
4 |
import sys
|