SebastianBodza
commited on
Commit
·
a748a2b
1
Parent(s):
d36cc58
Update README.md
Browse files
README.md
CHANGED
@@ -3,6 +3,7 @@
|
|
3 |
|
4 |
Multi-GPU bugfix for MPT-30B
|
5 |
|
|
|
6 |
|
7 |
|
8 |
This is the Python model code for MPT-7B patched so that it can be used with a LoRA. Note that while I tested that it works and I get reasonable results out, it is very possible that the model isn't being trained correctly. The model code specifically says that left padding is not supported, but I forcibly did so and got decent results.
|
|
|
3 |
|
4 |
Multi-GPU bugfix for MPT-30B
|
5 |
|
6 |
+
Patch based on: https://github.com/iwalton3/mpt-lora-patch
|
7 |
|
8 |
|
9 |
This is the Python model code for MPT-7B patched so that it can be used with a LoRA. Note that while I tested that it works and I get reasonable results out, it is very possible that the model isn't being trained correctly. The model code specifically says that left padding is not supported, but I forcibly did so and got decent results.
|