Instructions to use grimjim/madwind-wizard-7B with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use grimjim/madwind-wizard-7B with Transformers:

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

pipe = pipeline("text-generation", model="grimjim/madwind-wizard-7B")

# Load model directly
from transformers import AutoTokenizer, AutoModelForCausalLM

tokenizer = AutoTokenizer.from_pretrained("grimjim/madwind-wizard-7B")
model = AutoModelForCausalLM.from_pretrained("grimjim/madwind-wizard-7B")

Notebooks
Google Colab
Kaggle
Local Apps

vLLM

How to use grimjim/madwind-wizard-7B with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "grimjim/madwind-wizard-7B"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "grimjim/madwind-wizard-7B",
		"prompt": "Once upon a time,",
		"max_tokens": 512,
		"temperature": 0.5
	}'

Use Docker

docker model run hf.co/grimjim/madwind-wizard-7B

SGLang

How to use grimjim/madwind-wizard-7B 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 "grimjim/madwind-wizard-7B" \
    --host 0.0.0.0 \
    --port 30000
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:30000/v1/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "grimjim/madwind-wizard-7B",
		"prompt": "Once upon a time,",
		"max_tokens": 512,
		"temperature": 0.5
	}'

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 "grimjim/madwind-wizard-7B" \
        --host 0.0.0.0 \
        --port 30000
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:30000/v1/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "grimjim/madwind-wizard-7B",
		"prompt": "Once upon a time,",
		"max_tokens": 512,
		"temperature": 0.5
	}'

Docker Model Runner
How to use grimjim/madwind-wizard-7B with Docker Model Runner:
```
docker model run hf.co/grimjim/madwind-wizard-7B
```
Browse Quantizations to use this model in llama.cpp, Ollama, LM Studio, or any compatible app.

madwind-wizard-7B

This is a merge of pre-trained 7B language models created using mergekit.

The intended goal of this merge was to combine the 32K context window of Mistral v0.2 base with the richness and strength of the Zephyr Beta and WizardLM 2 models. This was a mixed-precision merge, promoting Mistral v0.2 base from fp16 to bf16.

The result can be used for text generation. Note that Zephyr Beta training removed in-built alignment from datasets, resulting in a model more likely to generate problematic text when prompted. This merge appears to have inherited that feature.

Full weights: grimjim/madwind-wizard-7B
GGUF quants: grimjim/madwind-wizard-7B-GGUF

Merge Details

Merge Method

This model was merged using the SLERP merge method.

Models Merged

The following models were included in the merge:

Configuration

The following YAML configuration was used to produce this model:

slices:
  - sources:
    - model: alpindale/Mistral-7B-v0.2-hf
      layer_range: [0,32]
    - model: grimjim/zephyr-beta-wizardLM-2-merge-7B
      layer_range: [0,32]
merge_method: slerp
base_model: alpindale/Mistral-7B-v0.2-hf
parameters:
  t:
    - value: 0.5
dtype: bfloat16