Instructions to use junzzhu/atoMixtral-58K-5x5-DigitMesh with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use junzzhu/atoMixtral-58K-5x5-DigitMesh with Transformers:

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

pipe = pipeline("text-generation", model="junzzhu/atoMixtral-58K-5x5-DigitMesh")

# Load model directly
from transformers import AutoTokenizer, AutoModelForCausalLM

tokenizer = AutoTokenizer.from_pretrained("junzzhu/atoMixtral-58K-5x5-DigitMesh")
model = AutoModelForCausalLM.from_pretrained("junzzhu/atoMixtral-58K-5x5-DigitMesh", device_map="auto")

Notebooks
Google Colab
Kaggle
Local Apps Settings

vLLM

How to use junzzhu/atoMixtral-58K-5x5-DigitMesh with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "junzzhu/atoMixtral-58K-5x5-DigitMesh"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "junzzhu/atoMixtral-58K-5x5-DigitMesh",
		"prompt": "Once upon a time,",
		"max_tokens": 512,
		"temperature": 0.5
	}'

Use Docker

docker model run hf.co/junzzhu/atoMixtral-58K-5x5-DigitMesh

SGLang

How to use junzzhu/atoMixtral-58K-5x5-DigitMesh 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 "junzzhu/atoMixtral-58K-5x5-DigitMesh" \
    --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": "junzzhu/atoMixtral-58K-5x5-DigitMesh",
		"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 "junzzhu/atoMixtral-58K-5x5-DigitMesh" \
        --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": "junzzhu/atoMixtral-58K-5x5-DigitMesh",
		"prompt": "Once upon a time,",
		"max_tokens": 512,
		"temperature": 0.5
	}'

Docker Model Runner
How to use junzzhu/atoMixtral-58K-5x5-DigitMesh with Docker Model Runner:
```
docker model run hf.co/junzzhu/atoMixtral-58K-5x5-DigitMesh
```

atoMixtral-58K-5x5-DigitMesh / README.md

junzzhu

Update README.md

e2e2949 verified 7 months ago

preview code

Raw

History Blame Contribute Delete

1.96 kB

	---
	language:
	- en
	tags:
	- mixtral
	- MixtralForCausalLM
	- mixture-of-experts
	- moe
	- digit-recognition
	- pattern-recognition
	- toy-model
	- educational
	license: apache-2.0
	library_name: transformers
	pipeline_tag: text-generation
	---

	# AtoMixtral-58K-5x5-DigitMesh

	A minimal 58K parameter Mixture-of-Experts (MoE) model for 5×5 digit mesh recognition, built on the MixtralForCausalLM architecture.

	## Model Description

	AtoMixtral-58K-5x5-DigitMesh is an ultra-lightweight MoE causal language model for efficient digit recognition from 5×5 binary mesh patterns. With only 58K parameters and 2 experts, this "atom-sized" MoE model demonstrates effective pattern recognition with sparse expert activation.

	### Key Specifications

	- Architecture: MixtralForCausalLM (Mixture-of-Experts)
	- Parameters: ~58K
	- Experts: 2 local experts, 1 active per token
	- Input: 5×5 binary mesh (25 tokens)
	- Output: Digit tokens (D0-D9)
	- Vocabulary Size: 14 tokens
	- Context Length: 32 tokens
	- Hidden Size: 32, Layers: 2, Attention Heads: 4

	## Quick Start

	### Serving with vLLM

	```bash
	python -m vllm.entrypoints.openai.api_server \
	--model junzzhu/atoMixtral-58K-5x5-DigitMesh \
	--max-model-len 32
	```

	### Test Example

	```bash
	curl http://localhost:8000/v1/completions \
	-H 'Content-Type: application/json' \
	-d '{
	"model": "junzzhu/atoMixtral-58K-5x5-DigitMesh",
	"prompt": "1 1 1 1 1 0 0 0 0 1 0 0 0 1 0 0 0 1 0 0 0 1 0 0 0 <SEP>",
	"max_tokens": 1,
	"temperature": 0
	}'
	```
	Expected output: `D7`

	## Input Format

	25 space-separated binary values (0 or 1) representing a 5×5 grid, followed by `<SEP>`:

	```
	[5 values] [5 values] [5 values] [5 values] [5 values] <SEP>
	```

	## Use Cases

	- MoE architecture research at minimal scale
	- Educational demonstrations of sparse expert models
	- Resource-constrained digit recognition
	- Pattern recognition proof-of-concepts

	## License

	Apache-2.0