Instructions to use starble-dev/Hollow-Tail-V1-12B with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use starble-dev/Hollow-Tail-V1-12B with Transformers:

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

pipe = pipeline("text-generation", model="starble-dev/Hollow-Tail-V1-12B")
messages = [
    {"role": "user", "content": "Who are you?"},
]
pipe(messages)

# Load model directly
from transformers import AutoTokenizer, AutoModelForCausalLM

tokenizer = AutoTokenizer.from_pretrained("starble-dev/Hollow-Tail-V1-12B")
model = AutoModelForCausalLM.from_pretrained("starble-dev/Hollow-Tail-V1-12B")

Notebooks
Google Colab
Kaggle
Local Apps

vLLM

How to use starble-dev/Hollow-Tail-V1-12B with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "starble-dev/Hollow-Tail-V1-12B"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "starble-dev/Hollow-Tail-V1-12B",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Use Docker

docker model run hf.co/starble-dev/Hollow-Tail-V1-12B

SGLang

How to use starble-dev/Hollow-Tail-V1-12B 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 "starble-dev/Hollow-Tail-V1-12B" \
    --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": "starble-dev/Hollow-Tail-V1-12B",
		"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 "starble-dev/Hollow-Tail-V1-12B" \
        --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": "starble-dev/Hollow-Tail-V1-12B",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Docker Model Runner
How to use starble-dev/Hollow-Tail-V1-12B with Docker Model Runner:
```
docker model run hf.co/starble-dev/Hollow-Tail-V1-12B
```
Browse Quantizations to use this model in llama.cpp, Ollama, LM Studio, or any compatible app.

General Use Sampling:
Mistral-Nemo-12B is very sensitive to the temperature sampler, try values near 0.3 at first or else you will get some weird results. This is mentioned by MistralAI at their Transformers section.

Best Samplers:
I found best success using the following for Hollow-Tail-V1-12B:
Temperature: 1.2
Top K: -1
Min P: 0.05
Rep Penalty: 1.08

Results

Disclaimer: This is a model merge!

Seems to be a bit smarter than I expected from my experience. May need a bit of guidance through system prompts at the beginning but it was quite fun to use. Though one thing I've realized with Mistral-Nemo is that the model does not really seem to end correctly, so instead I use the following format:

Obviously this is just my personal experience but I find this to be a good setup, I strongly recommend you experiment with different system formats and see which is better for you. Note that none of these models are fine-tuned for this specific format, I believe most of them are fine-tuned on Mistral's original [INST] and [/INST] format or ChatML.

<[start_system]>
You are a professional writer.
<[STOP]>
<[start_prompt]>
User prompt here.
<[STOP]>
<[start_model]>
Model response here.
<[STOP]>

Original Models:

Sao10K/MN-12B-Lyra-v2a1 (Thank you so much for your work ♥)
migtissera/Tess-3-Mistral-Nemo-12B (Thank you so much for your work ♥)
TheDrummer/Rocinante-12B-v1.1 (Thank you so much for your work ♥)

GGUF Quants:

starble-dev/Hollow-Tail-V1-12B-GGUF

Original Model Licenses:

Sao10K/MN-12B-Lyra-v2a1 is licensed under Creative Commons Attribution Non Commercial 4.0
migtissera/Tess-3-Mistral-Nemo-12B is licensed under apache-2.0
TheDrummer/Rocinante-12B-v1.1 license is not specified

Hollow-Tail-V1-12B

This is a merge of pre-trained language models created using mergekit.

Merge Details

Merge Method

This model was merged using the linear merge method using models/Rocinante-12B-v1.1 as a base.

Models Merged

The following models were included in the merge:

migtissera/Tess-3-Mistral-Nemo-12B
Sao10K/MN-12B-Lyra-v2a1
TheDrummer/Rocinante-12B-v1.1

Configuration

The following YAML configuration was used to produce this model:

models:
    - model: Sao10K/MN-12B-Lyra-v2a1
      parameters:
        weight: 0.8
    - model: migtissera/Tess-3-Mistral-Nemo-12B
      parameters:
        weight: 0.2
    - model: TheDrummer/Rocinante-12B-v1.1
      parameters:
        weight: 0.8
merge_method: linear
base_model: TheDrummer/Rocinante-12B-v1.1
parameters:
    normalize: true
    int8_mask: true
dtype: bfloat16