Update README.md
Browse files
README.md
CHANGED
|
@@ -19,6 +19,8 @@ library_name: transformers
|
|
| 19 |
|
| 20 |
# internlm2_5-20b-chat-abliterated
|
| 21 |
|
|
|
|
|
|
|
| 22 |
This is a new approach for abliterating models using CPU only. I was able to abliterate this model using free kaggle processing with no accelerator.
|
| 23 |
1. Obtain refusal direction vector using a quant model with llama.cpp (llama-cpp-python and ggml-python).
|
| 24 |
2. Orthogonalize each .safetensors files directly from original repo and upload to a new repo. (one at a time)
|
|
|
|
| 19 |
|
| 20 |
# internlm2_5-20b-chat-abliterated
|
| 21 |
|
| 22 |
+
## NOTE: This model is a WIP and subject to frequent changes over the next day or two, the model is being tested to find the best layer to use for refusal mitigation.
|
| 23 |
+
|
| 24 |
This is a new approach for abliterating models using CPU only. I was able to abliterate this model using free kaggle processing with no accelerator.
|
| 25 |
1. Obtain refusal direction vector using a quant model with llama.cpp (llama-cpp-python and ggml-python).
|
| 26 |
2. Orthogonalize each .safetensors files directly from original repo and upload to a new repo. (one at a time)
|