Instructions to use Knobi3/EvoMerge1 with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use Knobi3/EvoMerge1 with Transformers:

# Use a pipeline as a high-level helper
from transformers import pipeline

pipe = pipeline("text-generation", model="Knobi3/EvoMerge1")
messages = [
    {"role": "user", "content": "Who are you?"},
]
pipe(messages)

# Load model directly
from transformers import AutoTokenizer, AutoModelForCausalLM

tokenizer = AutoTokenizer.from_pretrained("Knobi3/EvoMerge1")
model = AutoModelForCausalLM.from_pretrained("Knobi3/EvoMerge1")
messages = [
    {"role": "user", "content": "Who are you?"},
]
inputs = tokenizer.apply_chat_template(
	messages,
	add_generation_prompt=True,
	tokenize=True,
	return_dict=True,
	return_tensors="pt",
).to(model.device)

outputs = model.generate(**inputs, max_new_tokens=40)
print(tokenizer.decode(outputs[0][inputs["input_ids"].shape[-1]:]))

Notebooks
Google Colab
Kaggle
Local Apps

vLLM

How to use Knobi3/EvoMerge1 with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "Knobi3/EvoMerge1"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "Knobi3/EvoMerge1",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Use Docker

docker model run hf.co/Knobi3/EvoMerge1

SGLang

How to use Knobi3/EvoMerge1 with SGLang:

Install from pip and serve model

# Install SGLang from pip:
pip install sglang
# Start the SGLang server:
python3 -m sglang.launch_server \
    --model-path "Knobi3/EvoMerge1" \
    --host 0.0.0.0 \
    --port 30000
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:30000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "Knobi3/EvoMerge1",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Use Docker images

docker run --gpus all \
    --shm-size 32g \
    -p 30000:30000 \
    -v ~/.cache/huggingface:/root/.cache/huggingface \
    --env "HF_TOKEN=<secret>" \
    --ipc=host \
    lmsysorg/sglang:latest \
    python3 -m sglang.launch_server \
        --model-path "Knobi3/EvoMerge1" \
        --host 0.0.0.0 \
        --port 30000
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:30000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "Knobi3/EvoMerge1",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Docker Model Runner
How to use Knobi3/EvoMerge1 with Docker Model Runner:
```
docker model run hf.co/Knobi3/EvoMerge1
```
Browse Quantizations to use this model in llama.cpp, Ollama, LM Studio, or any compatible app.

Evolutionary model merging

This is a merge of pre-trained language models created using mergekit. However, only 2 evalations.

Merge Details

Merge Method

This model was merged using the task arithmetic merge method using Mistral-7B-Instruct-v0.2 as a base.

Models Merged

The following models were included in the merge:

Starling-LM-7B-beta_581094980
NeuralBeagle14-7B_2368216670
Mistral-7B-Instruct-v0.2_674785087
sol_2748493923

Configuration

The following YAML configuration was used to produce this model:

base_model: /content/evol_merge_storage/input_models/Mistral-7B-Instruct-v0.2_674785087
dtype: bfloat16
merge_method: task_arithmetic
parameters:
  int8_mask: 1.0
  normalize: 0.0
slices:
- sources:
  - layer_range: [0, 8]
    model: /content/evol_merge_storage/input_models/NeuralBeagle14-7B_2368216670
    parameters:
      weight: 0.445198554642378
  - layer_range: [0, 8]
    model: /content/evol_merge_storage/input_models/Starling-LM-7B-beta_581094980
    parameters:
      weight: 0.005659554655209731
  - layer_range: [0, 8]
    model: /content/evol_merge_storage/input_models/sol_2748493923
    parameters:
      weight: 0.3149085451085008
  - layer_range: [0, 8]
    model: /content/evol_merge_storage/input_models/Mistral-7B-Instruct-v0.2_674785087
- sources:
  - layer_range: [8, 16]
    model: /content/evol_merge_storage/input_models/NeuralBeagle14-7B_2368216670
    parameters:
      weight: 0.07684631401806669
  - layer_range: [8, 16]
    model: /content/evol_merge_storage/input_models/Starling-LM-7B-beta_581094980
    parameters:
      weight: 0.4987353671063862
  - layer_range: [8, 16]
    model: /content/evol_merge_storage/input_models/sol_2748493923
    parameters:
      weight: 0.5082785378119663
  - layer_range: [8, 16]
    model: /content/evol_merge_storage/input_models/Mistral-7B-Instruct-v0.2_674785087
- sources:
  - layer_range: [16, 24]
    model: /content/evol_merge_storage/input_models/NeuralBeagle14-7B_2368216670
    parameters:
      weight: 0.5687639014384054
  - layer_range: [16, 24]
    model: /content/evol_merge_storage/input_models/Starling-LM-7B-beta_581094980
    parameters:
      weight: 0.5007180726769771
  - layer_range: [16, 24]
    model: /content/evol_merge_storage/input_models/sol_2748493923
    parameters:
      weight: 0.3855348742111532
  - layer_range: [16, 24]
    model: /content/evol_merge_storage/input_models/Mistral-7B-Instruct-v0.2_674785087
- sources:
  - layer_range: [24, 32]
    model: /content/evol_merge_storage/input_models/NeuralBeagle14-7B_2368216670
    parameters:
      weight: 0.17691410154696996
  - layer_range: [24, 32]
    model: /content/evol_merge_storage/input_models/Starling-LM-7B-beta_581094980
    parameters:
      weight: 0.2031085126199354
  - layer_range: [24, 32]
    model: /content/evol_merge_storage/input_models/sol_2748493923
    parameters:
      weight: 0.5603073995258283
  - layer_range: [24, 32]
    model: /content/evol_merge_storage/input_models/Mistral-7B-Instruct-v0.2_674785087

Downloads last month: 9

Safetensors

Model size

7B params

Tensor type

BF16

Model tree for Knobi3/EvoMerge1

Quantizations

1 model

Spaces using Knobi3/EvoMerge1 9

Paper for Knobi3/EvoMerge1

Editing Models with Task Arithmetic

Paper • 2212.04089 • Published Dec 8, 2022 • 8