Instructions to use crabby-devil/zils-mini-mica-4b with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use crabby-devil/zils-mini-mica-4b with Transformers:

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

pipe = pipeline("image-text-to-text", model="crabby-devil/zils-mini-mica-4b")

# pip install -U transformers accelerate
# Load model directly
from transformers import AutoProcessor, AutoModelForMultimodalLM

processor = AutoProcessor.from_pretrained("crabby-devil/zils-mini-mica-4b")
model = AutoModelForMultimodalLM.from_pretrained("crabby-devil/zils-mini-mica-4b", device_map="auto")

Notebooks
Google Colab
Kaggle
Local Apps Settings

vLLM

How to use crabby-devil/zils-mini-mica-4b with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "crabby-devil/zils-mini-mica-4b"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "crabby-devil/zils-mini-mica-4b",
		"prompt": "Once upon a time,",
		"max_tokens": 512,
		"temperature": 0.5
	}'

Use Docker

docker model run hf.co/crabby-devil/zils-mini-mica-4b

SGLang

How to use crabby-devil/zils-mini-mica-4b 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 "crabby-devil/zils-mini-mica-4b" \
    --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": "crabby-devil/zils-mini-mica-4b",
		"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 "crabby-devil/zils-mini-mica-4b" \
        --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": "crabby-devil/zils-mini-mica-4b",
		"prompt": "Once upon a time,",
		"max_tokens": 512,
		"temperature": 0.5
	}'

Docker Model Runner
How to use crabby-devil/zils-mini-mica-4b with Docker Model Runner:
```
docker model run hf.co/crabby-devil/zils-mini-mica-4b
```

Mica-v0.5-4B: Zils d3-mini soup body

This is an unmodified redistribution of the bf16 weights of sky7350/Mica-v0.5-4B at revision f87da85ba6f53c0a67c8b9407a03d2b18274c24a. It is a decision fine-tune of Qwen/Qwen3.5-4B.

The weights are published as a body for the Zils (Bittensor subnet 127) d3-mini soup pool.
The two bf16 shards, the shard index and config.json are byte-identical to the source revision.
The int8 copy under int8/ is not included.
LICENSE is the Apache-2.0 text.
zils-owner.json names the Bittensor hotkey that committed this body.

All credit for the model goes to its original authors; see the source model card. Licence: Apache-2.0, the same as the source.

Downloads last month: 46

Safetensors

Model size

5B params

Tensor type

BF16

F32

Model tree for crabby-devil/zils-mini-mica-4b

Base model

Qwen/Qwen3.5-4B-Base

Finetuned

Qwen/Qwen3.5-4B

Finetuned

sky7350/Mica-v0.5-4B

Finetuned

(1)

this model