Instructions to use grimjim/Llama-3-Instruct-abliteration-LoRA-8B with libraries, inference providers, notebooks, and local apps. Follow these links to get started.
- Libraries
- Transformers
How to use grimjim/Llama-3-Instruct-abliteration-LoRA-8B with Transformers:
# Load model directly from transformers import AutoModel model = AutoModel.from_pretrained("grimjim/Llama-3-Instruct-abliteration-LoRA-8B", dtype="auto") - PEFT
How to use grimjim/Llama-3-Instruct-abliteration-LoRA-8B with PEFT:
Task type is invalid.
- Notebooks
- Google Colab
- Kaggle
Initial release
Browse files- README.md +24 -0
- adapter_config.json +33 -0
- adapter_model.safetensors +3 -0
README.md
CHANGED
|
@@ -1,3 +1,27 @@
|
|
| 1 |
---
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 2 |
license: llama3
|
| 3 |
---
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
---
|
| 2 |
+
base_model:
|
| 3 |
+
- meta-llama/Meta-Llama-3-8B-Instruct
|
| 4 |
+
- failspy/Meta-Llama-3-8B-Instruct-abliterated-v3
|
| 5 |
+
library_name: transformers
|
| 6 |
+
tags:
|
| 7 |
+
- mergekit
|
| 8 |
+
- peft
|
| 9 |
license: llama3
|
| 10 |
---
|
| 11 |
+
# Llama-3-Instruct-abliteration-LoRA-8B
|
| 12 |
+
|
| 13 |
+
This is a rank=32 LoRA extracted from a language model. It was extracted using [mergekit](https://github.com/arcee-ai/mergekit).
|
| 14 |
+
|
| 15 |
+
This LoRA is derived from the refusal ablation vector also computed in [Llama-3-Instruct-abliteration-OVA-8B](https://huggingface.co/grimjim/Llama-3-Instruct-abliteration-OVA-8B), and is around 5 orders of magnitude smaller than full bf16 weights of the base model.
|
| 16 |
+
|
| 17 |
+
## LoRA Details
|
| 18 |
+
|
| 19 |
+
This LoRA adapter was extracted from [failspy/Meta-Llama-3-8B-Instruct-abliterated-v3](https://huggingface.co/failspy/Meta-Llama-3-8B-Instruct-abliterated-v3) and uses [meta-llama/Meta-Llama-3-8B-Instruct](https://huggingface.co/meta-llama/Meta-Llama-3-8B-Instruct) as a base.
|
| 20 |
+
|
| 21 |
+
### Parameters
|
| 22 |
+
|
| 23 |
+
The following command was used to extract this LoRA adapter:
|
| 24 |
+
|
| 25 |
+
```sh
|
| 26 |
+
mergekit-extract-lora meta-llama/Meta-Llama-3-8B-Instruct failspy/Meta-Llama-3-8B-Instruct-abliterated-v3 OUTPUT_PATH --rank=32 --model_name=Llama-3-8B-Instruct-counter-refusal
|
| 27 |
+
```
|
adapter_config.json
ADDED
|
@@ -0,0 +1,33 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"alpha_pattern": {},
|
| 3 |
+
"auto_mapping": null,
|
| 4 |
+
"base_model_name_or_path": "meta-llama/Meta-Llama-3-8B-Instruct",
|
| 5 |
+
"bias": "none",
|
| 6 |
+
"fan_in_fan_out": false,
|
| 7 |
+
"inference_mode": true,
|
| 8 |
+
"init_lora_weights": true,
|
| 9 |
+
"layers_pattern": null,
|
| 10 |
+
"layers_to_transform": null,
|
| 11 |
+
"loftq_config": {},
|
| 12 |
+
"lora_alpha": 32,
|
| 13 |
+
"lora_dropout": 0,
|
| 14 |
+
"megatron_config": null,
|
| 15 |
+
"megatron_core": "megatron.core",
|
| 16 |
+
"modules_to_save": null,
|
| 17 |
+
"peft_type": "LORA",
|
| 18 |
+
"r": 32,
|
| 19 |
+
"rank_pattern": {},
|
| 20 |
+
"revision": null,
|
| 21 |
+
"target_modules": [
|
| 22 |
+
"gate_proj",
|
| 23 |
+
"k_proj",
|
| 24 |
+
"up_proj",
|
| 25 |
+
"v_proj",
|
| 26 |
+
"o_proj",
|
| 27 |
+
"down_proj",
|
| 28 |
+
"q_proj",
|
| 29 |
+
"lm_head"
|
| 30 |
+
],
|
| 31 |
+
"task_type": "CAUSAL_LM",
|
| 32 |
+
"use_rslora": false
|
| 33 |
+
}
|
adapter_model.safetensors
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:0f1d4918d10fb6f600f731ec467245ab2000c6fe176644de9650bb45faa8dfcb
|
| 3 |
+
size 176303496
|