Neo_7b-merge2
Neo_7b-merge2 is a merge of the following models using LazyMergekit:
🧩 Configuration
slices:
# Group 1
- sources:
- model: m-a-p/neo_7b
layer_range: [0, 0]
- model: m-a-p/neo_7b
layer_range: [3, 3]
- sources:
- model: m-a-p/neo_7b
layer_range: [1, 1]
- model: m-a-p/neo_7b
layer_range: [3, 3]
- sources:
- model: m-a-p/neo_7b
layer_range: [2, 2]
- model: m-a-p/neo_7b
layer_range: [3, 3]
# Group 2
- sources:
- model: m-a-p/neo_7b
layer_range: [4, 4]
- model: m-a-p/neo_7b
layer_range: [7, 7]
- sources:
- model: m-a-p/neo_7b
layer_range: [5, 5]
- model: m-a-p/neo_7b
layer_range: [7, 7]
- sources:
- model: m-a-p/neo_7b
layer_range: [6, 6]
- model: m-a-p/neo_7b
layer_range: [7, 7]
# Group 3
- sources:
- model: m-a-p/neo_7b
layer_range: [8, 8]
- model: m-a-p/neo_7b
layer_range: [11, 11]
- sources:
- model: m-a-p/neo_7b
layer_range: [9, 9]
- model: m-a-p/neo_7b
layer_range: [11, 11]
- sources:
- model: m-a-p/neo_7b
layer_range: [10, 10]
- model: m-a-p/neo_7b
layer_range: [11, 11]
# Group 4
- sources:
- model: m-a-p/neo_7b
layer_range: [12, 12]
- model: m-a-p/neo_7b
layer_range: [15, 15]
- sources:
- model: m-a-p/neo_7b
layer_range: [13, 13]
- model: m-a-p/neo_7b
layer_range: [15, 15]
- sources:
- model: m-a-p/neo_7b
layer_range: [14, 14]
- model: m-a-p/neo_7b
layer_range: [15, 15]
# Group 5
- sources:
- model: m-a-p/neo_7b
layer_range: [16, 16]
- model: m-a-p/neo_7b
layer_range: [19, 19]
- sources:
- model: m-a-p/neo_7b
layer_range: [17, 17]
- model: m-a-p/neo_7b
layer_range: [19, 19]
- sources:
- model: m-a-p/neo_7b
layer_range: [18, 18]
- model: m-a-p/neo_7b
layer_range: [19, 19]
# Group 6
- sources:
- model: m-a-p/neo_7b
layer_range: [20, 20]
- model: m-a-p/neo_7b
layer_range: [23, 23]
- sources:
- model: m-a-p/neo_7b
layer_range: [21, 21]
- model: m-a-p/neo_7b
layer_range: [23, 23]
- sources:
- model: m-a-p/neo_7b
layer_range: [22, 22]
- model: m-a-p/neo_7b
layer_range: [23, 23]
# Group 7 (last group)
- sources:
- model: m-a-p/neo_7b
layer_range: [24, 24]
- model: m-a-p/neo_7b
layer_range: [27, 27]
- sources:
- model: m-a-p/neo_7b
layer_range: [25, 25]
- model: m-a-p/neo_7b
layer_range: [27, 27]
- sources:
- model: m-a-p/neo_7b
layer_range: [26, 26]
- model: m-a-p/neo_7b
layer_range: [27, 27]
merge_method: slerp
base_model: m-a-p/neo_7b
parameters:
t: 0.3333 # Apply 1/3 of the 4th layer to each of the previous 3 layers
dtype: bfloat16
output_path: ./merged_redistributed_neo_7b
💻 Usage
!pip install -qU transformers accelerate
from transformers import AutoTokenizer
import transformers
import torch
model = "DewEfresh/Neo_7b-merge2"
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"])