Instructions to use frameai/Loxa-3B with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use frameai/Loxa-3B with Transformers:

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

pipe = pipeline("text-generation", model="frameai/Loxa-3B")
messages = [
    {"role": "user", "content": "Who are you?"},
]
pipe(messages)

# Load model directly
from transformers import AutoTokenizer, AutoModelForCausalLM

tokenizer = AutoTokenizer.from_pretrained("frameai/Loxa-3B")
model = AutoModelForCausalLM.from_pretrained("frameai/Loxa-3B")
messages = [
    {"role": "user", "content": "Who are you?"},
]
inputs = tokenizer.apply_chat_template(
	messages,
	add_generation_prompt=True,
	tokenize=True,
	return_dict=True,
	return_tensors="pt",
).to(model.device)

outputs = model.generate(**inputs, max_new_tokens=40)
print(tokenizer.decode(outputs[0][inputs["input_ids"].shape[-1]:]))

Notebooks
Google Colab
Kaggle
Local Apps

vLLM

How to use frameai/Loxa-3B with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "frameai/Loxa-3B"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "frameai/Loxa-3B",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Use Docker

docker model run hf.co/frameai/Loxa-3B

SGLang

How to use frameai/Loxa-3B 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 "frameai/Loxa-3B" \
    --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": "frameai/Loxa-3B",
		"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 "frameai/Loxa-3B" \
        --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": "frameai/Loxa-3B",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Docker Model Runner
How to use frameai/Loxa-3B with Docker Model Runner:
```
docker model run hf.co/frameai/Loxa-3B
```
Browse Quantizations to use this model in llama.cpp, Ollama, LM Studio, or any compatible app.

Model Card for Loxa

Model Name: Loxa-3B

Model Family: Loxa

Creator: AIFRAME

Description: Loxa-3B is a powerful language model designed for optimal performance on CPU resources, particularly Raspberry Pi 4 and 5 (8GB+ RAM). It excels in math, code, chat, help, science, and formal conversations, achieving 92% total accuracy.

Capabilities:

Mathematics: Solves problems, performs calculations, explains concepts.
Code: Generates code, understands/debugs existing code, provides explanations.
Chat: Engages in conversations, provides informative and helpful responses.
Help: Offers guidance and clear explanations across various topics.
Science: Discusses scientific topics, explains phenomena, provides insights.
Formal Conversations: Maintains formal etiquette and respectful language.

Performance:

Accuracy: 92% total accuracy.
Resource Usage: Optimized for Raspberry Pi 4/5 (8GB+ RAM). Consult documentation for detailed metrics.

Intended Use: Educational purposes, personal projects, embedded systems, resource-constrained environments.

Limitations:

May produce incorrect or nonsensical outputs. Exercise caution for critical tasks. Performance may be affected by input complexity/length. See documentation for details on limitations and biases.

How to Use: See accompanying documentation for installation and usage instructions.

Code Example:

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

messages = [
    {"role": "user", "content": "Who are you?"},
]
pipe = pipeline("text-generation", model="explorewithai/Loxa-3B") # Using 'explorewithai' as a placeholder organization
result = pipe(messages)
print(result)

Downloads last month: 4

Model tree for frameai/Loxa-3B

Unable to build the model tree, the base model loops to the model itself. Learn more.

Space using frameai/Loxa-3B 1

Collection including frameai/Loxa-3B

Loxa

Collection

a Loxa family models are best models to running on CPU and GPU with high quality(=>92% accuracy) • 5 items • Updated Feb 3, 2025 • 2