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

Libraries

How to use flwrlabs/Lizzy-7B with Transformers:

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

pipe = pipeline("text-generation", model="flwrlabs/Lizzy-7B", trust_remote_code=True)
messages = [
    {"role": "user", "content": "Who are you?"},
]
pipe(messages)

# pip install -U transformers accelerate
# Load model directly
from transformers import AutoModelForCausalLM
model = AutoModelForCausalLM.from_pretrained("flwrlabs/Lizzy-7B", trust_remote_code=True, device_map="auto")

Notebooks
Google Colab
Kaggle
Local Apps Settings

vLLM

How to use flwrlabs/Lizzy-7B with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "flwrlabs/Lizzy-7B"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "flwrlabs/Lizzy-7B",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Use Docker

docker model run hf.co/flwrlabs/Lizzy-7B

SGLang

How to use flwrlabs/Lizzy-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 "flwrlabs/Lizzy-7B" \
    --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": "flwrlabs/Lizzy-7B",
		"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 "flwrlabs/Lizzy-7B" \
        --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": "flwrlabs/Lizzy-7B",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Docker Model Runner
How to use flwrlabs/Lizzy-7B with Docker Model Runner:
```
docker model run hf.co/flwrlabs/Lizzy-7B
```

First MLX port

by Strikesure5555 - opened Sep 4

Discussion

Strikesure5555

Sep 4

•

edited Sep 4

There's no MLX/Apple Silicon support for this model anywhere, at least not that I could find (I accept the possibility of pilot error — probably why I am not a pilot). The official GGUFs run only through Flower Labs' own llama.cpp fork, since the architecture isn't in mainline llama.cpp. I built a native MLX port instead: real architecture implementation, not a trust_remote_code workaround.

One thing worth knowing if you're loading the original repo directly: trust_remote_code requires transformers>=5.4.0 — works fine from there through current, but breaks on anything older (tokenizer/cache-API issues below 5.4).

Validated against the reference PyTorch implementation: fp32 layerwise parity, bf16 agreement, greedy-decode exact match, KV-cache self-consistency. Plus negative controls, confirming the checks actually catch a wrong implementation.

Weights:

bf16 (reference): https://huggingface.co/Strikesure5555/Lizzy-7B-mlx-bf16
8-bit: https://huggingface.co/Strikesure5555/Lizzy-7B-mlx-8bit
4-bit: https://huggingface.co/Strikesure5555/Lizzy-7B-mlx-4bit
Preparing an mlx-lm PR to add native support upstream.

Upload images, audio, and videos by dragging in the text input, pasting, or clicking here.

Tap or paste here to upload images

· Sign up or log in to comment