update on breaking changes
Browse files
README.md
CHANGED
@@ -7,6 +7,15 @@ This LoRA trained for 3 epochs and has been converted to int4 via GPTQ method. S
|
|
7 |
|
8 |
https://github.com/qwopqwop200/GPTQ-for-LLaMa
|
9 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
10 |
# Update 2023-03-27
|
11 |
New weights have been added. The old .pt version is no longer supported and has been replaced by a 128 groupsize safetensors file. Update to the latest GPTQ to use it.
|
12 |
|
|
|
7 |
|
8 |
https://github.com/qwopqwop200/GPTQ-for-LLaMa
|
9 |
|
10 |
+
# Update 2023-04-03
|
11 |
+
Recent GPTQ commits have introduced breaking changes to model loading and you should use commit `a6f363e3f93b9fb5c26064b5ac7ed58d22e3f773` in the `cuda` branch.
|
12 |
+
|
13 |
+
If you're not familiar with the Git process
|
14 |
+
1. `git checkout a6f363e3f93b9fb5c26064b5ac7ed58d22e3f773`
|
15 |
+
2. `git switch -c cuda-stable`
|
16 |
+
|
17 |
+
This creates and switches to a `cuda-stable` branch to continue using the quantized models.
|
18 |
+
|
19 |
# Update 2023-03-27
|
20 |
New weights have been added. The old .pt version is no longer supported and has been replaced by a 128 groupsize safetensors file. Update to the latest GPTQ to use it.
|
21 |
|