You can deploy this model and start earning money today!

RichardErkhov/
natong19_-_Qwen2-7B-Instruct-abliterated-gguf

Name: natong19_-_Qwen2-7B-Instruct-abliterated-gguf
Availability: InStock

Views

No views yet

Quantization made by Richard Erkhov.

Github

Discord

Request more models

Qwen2-7B-Instruct-abliterated - GGUF

Model creator: https://huggingface.co/natong19/
Original model: https://huggingface.co/natong19/Qwen2-7B-Instruct-abliterated/

Name	Quant method	Size
Qwen2-7B-Instruct-abliterated.Q2_K.gguf	Q2_K	2.81GB
Qwen2-7B-Instruct-abliterated.IQ3_XS.gguf	IQ3_XS	3.12GB
Qwen2-7B-Instruct-abliterated.IQ3_S.gguf	IQ3_S	3.26GB
Qwen2-7B-Instruct-abliterated.Q3_K_S.gguf	Q3_K_S	3.25GB
Qwen2-7B-Instruct-abliterated.IQ3_M.gguf	IQ3_M	3.33GB
Qwen2-7B-Instruct-abliterated.Q3_K.gguf	Q3_K	3.55GB
Qwen2-7B-Instruct-abliterated.Q3_K_M.gguf	Q3_K_M	3.55GB
Qwen2-7B-Instruct-abliterated.Q3_K_L.gguf	Q3_K_L	3.81GB
Qwen2-7B-Instruct-abliterated.IQ4_XS.gguf	IQ4_XS	3.96GB
Qwen2-7B-Instruct-abliterated.Q4_0.gguf	Q4_0	4.13GB
Qwen2-7B-Instruct-abliterated.IQ4_NL.gguf	IQ4_NL	4.16GB
Qwen2-7B-Instruct-abliterated.Q4_K_S.gguf	Q4_K_S	4.15GB
Qwen2-7B-Instruct-abliterated.Q4_K.gguf	Q4_K	4.36GB
Qwen2-7B-Instruct-abliterated.Q4_K_M.gguf	Q4_K_M	4.36GB
Qwen2-7B-Instruct-abliterated.Q4_1.gguf	Q4_1	4.54GB
Qwen2-7B-Instruct-abliterated.Q5_0.gguf	Q5_0	4.95GB
Qwen2-7B-Instruct-abliterated.Q5_K_S.gguf	Q5_K_S	4.95GB
Qwen2-7B-Instruct-abliterated.Q5_K.gguf	Q5_K	5.07GB
Qwen2-7B-Instruct-abliterated.Q5_K_M.gguf	Q5_K_M	5.07GB
Qwen2-7B-Instruct-abliterated.Q5_1.gguf	Q5_1	5.36GB
Qwen2-7B-Instruct-abliterated.Q6_K.gguf	Q6_K	5.82GB
Qwen2-7B-Instruct-abliterated.Q8_0.gguf	Q8_0	7.54GB

Original model description:

license: apache-2.0 language:

en pipeline_tag: text-generation tags:
chat

Qwen2-7B-Instruct-abliterated

Introduction

Abliterated version of Qwen2-7B-Instruct using failspy's notebook. The model's strongest refusal directions have been ablated via weight orthogonalization, but the model may still refuse your request, misunderstand your intent, or provide unsolicited advice regarding ethics or safety.

Quickstart

python

from transformers import AutoModelForCausalLM, AutoTokenizer

model_id = "natong19/Qwen2-7B-Instruct-abliterated"
device = "cuda" # the device to load the model onto

model = AutoModelForCausalLM.from_pretrained(
    model_id,
    torch_dtype="auto",
    device_map="auto"
)
tokenizer = AutoTokenizer.from_pretrained(model_id)

prompt = "Give me a short introduction to large language model."
messages = [
    {"role": "system", "content": "You are a helpful assistant."},
    {"role": "user", "content": prompt}
]
text = tokenizer.apply_chat_template(
    messages,
    tokenize=False,
    add_generation_prompt=True
)
model_inputs = tokenizer([text], return_tensors="pt").to(device)

generated_ids = model.generate(
    model_inputs.input_ids,
    max_new_tokens=256
)
generated_ids = [
    output_ids[len(input_ids):] for input_ids, output_ids in zip(model_inputs.input_ids, generated_ids)
]

response = tokenizer.batch_decode(generated_ids, skip_special_tokens=True)[0]
print(response)

Evaluation

Evaluation framework: lm-evaluation-harness 0.4.2

Datasets	Qwen2-7B-Instruct	Qwen2-7B-Instruct-abliterated
ARC (25-shot)	62.5	62.5
GSM8K (5-shot)	73.0	72.2
HellaSwag (10-shot)	81.8	81.7
MMLU (5-shot)	70.7	70.5
TruthfulQA (0-shot)	57.3	55.0
Winogrande (5-shot)	76.2	77.4

RichardErkhov/natong19_-_Qwen2-7B-Instruct-abliterated-gguf