Extracted LoRA (mergekit)
Collection
PEFT-compatible LoRA adapters produced by mergekit-extract-lora • 17 items • Updated • 4
How to use ngxson/LoRA-Qwen2.5-72B-Instruct-abliterated with Transformers:
# Load model directly
from transformers import AutoModel
model = AutoModel.from_pretrained("ngxson/LoRA-Qwen2.5-72B-Instruct-abliterated", dtype="auto")How to use ngxson/LoRA-Qwen2.5-72B-Instruct-abliterated with PEFT:
Task type is invalid.
This is a LoRA extracted from a language model. It was extracted using mergekit.
This LoRA adapter was extracted from huihui-ai/Qwen2.5-72B-Instruct-abliterated and uses Qwen/Qwen2.5-72B-Instruct as a base.
The following command was used to extract this LoRA adapter:
mergekit-extract-lora huihui-ai/Qwen2.5-72B-Instruct-abliterated Qwen/Qwen2.5-72B-Instruct OUTPUT_PATH --rank=32 --device=cpu