metadata

base_model:
  - failspy/Meta-Llama-3-8B-Instruct-abliterated-v3
  - failspy/Meta-Llama-3-8B-Instruct-abliterated-v3
  - failspy/Meta-Llama-3-8B-Instruct-abliterated-v3
  - failspy/Meta-Llama-3-8B-Instruct-abliterated-v3
  - failspy/Meta-Llama-3-8B-Instruct-abliterated-v3
  - failspy/Meta-Llama-3-8B-Instruct-abliterated-v3
  - failspy/Meta-Llama-3-8B-Instruct-abliterated-v3
  - failspy/Meta-Llama-3-8B-Instruct-abliterated-v3
tags:
  - merge
  - mergekit
  - lazymergekit
  - failspy/Meta-Llama-3-8B-Instruct-abliterated-v3

experiment-9

experiment-9 is a merge of the following models using LazyMergekit:

🧩 Configuration

slices:
- sources:
  - layer_range: [0, 4]
    model: failspy/Meta-Llama-3-8B-Instruct-abliterated-v3
- sources:
  - layer_range: [5, 8]
    model: failspy/Meta-Llama-3-8B-Instruct-abliterated-v3
- sources:
  - layer_range: [9, 12]
    model: failspy/Meta-Llama-3-8B-Instruct-abliterated-v3
- sources:
  - layer_range: [13, 16]
    model: failspy/Meta-Llama-3-8B-Instruct-abliterated-v3
- sources:
  - layer_range: [17, 20]
    model: failspy/Meta-Llama-3-8B-Instruct-abliterated-v3
- sources:
  - layer_range: [21, 24]
    model: failspy/Meta-Llama-3-8B-Instruct-abliterated-v3
- sources:
  - layer_range: [25, 28]
    model: failspy/Meta-Llama-3-8B-Instruct-abliterated-v3
- sources:
  - layer_range: [29, 32]
    model: failspy/Meta-Llama-3-8B-Instruct-abliterated-v3
merge_method: passthrough
dtype: float16

💻 Usage

!pip install -qU transformers accelerate

from transformers import AutoTokenizer
import transformers
import torch

model = "KingNish/experiment-9"
messages = [{"role": "user", "content": "What is a large language model?"}]

tokenizer = AutoTokenizer.from_pretrained(model)
prompt = tokenizer.apply_chat_template(messages, tokenize=False, add_generation_prompt=True)
pipeline = transformers.pipeline(
    "text-generation",
    model=model,
    torch_dtype=torch.float16,
    device_map="auto",
)

outputs = pipeline(prompt, max_new_tokens=256, do_sample=True, temperature=0.7, top_k=50, top_p=0.95)
print(outputs[0]["generated_text"])