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

Libraries

How to use frameai/LoxaPro with Transformers:

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

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

# Load model directly
from transformers import AutoTokenizer, AutoModelForCausalLM

tokenizer = AutoTokenizer.from_pretrained("frameai/LoxaPro")
model = AutoModelForCausalLM.from_pretrained("frameai/LoxaPro")
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/LoxaPro with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "frameai/LoxaPro"
# 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/LoxaPro",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Use Docker

docker model run hf.co/frameai/LoxaPro

SGLang

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

Docker Model Runner
How to use frameai/LoxaPro with Docker Model Runner:
```
docker model run hf.co/frameai/LoxaPro
```

Loxa Pro

Model Description

Loxa Pro is a high-quality Large Language Model (LLM) designed to excel in a wide range of tasks, including:

Mathematics: Solving mathematical problems, performing calculations, and providing mathematical explanations.
Coding: Generating code snippets, debugging code, and answering programming-related questions.
General Questions: Answering a wide variety of general knowledge questions accurately and comprehensively.
Science Questions: Providing information and explanations on various scientific topics.
Daily Tasks: Assisting with everyday tasks, such as writing emails, setting reminders, generating to-do lists, and more.

Loxa Pro employs the CA (Combine Architectures) method, which enables it to effectively address diverse queries and tasks. This model surpasses its predecessors, Loxa-4B and Loxa-3B, in terms of accuracy and performance.

Key Features

High Accuracy: Loxa Pro demonstrates superior accuracy compared to Loxa-4B and Loxa-3B, providing more reliable and precise responses.
Broad Capabilities: Handles a diverse range of tasks, from complex mathematical problems and coding challenges to general knowledge and everyday tasks.
Optimized for Efficiency: The model is well-optimized to run efficiently even on smaller GPUs, making it accessible for users with limited computational resources.
CA (Combine Architectures) Method: Leverages the CA method to effectively combine different architectural strengths, enhancing overall performance.

Intended Use

Loxa Pro is intended for a wide range of applications, including:

Research: As a tool for research in natural language processing, artificial intelligence, and related fields.
Education: As an educational aid for students and educators in various subjects.
Development: As a component in building intelligent applications, chatbots, and virtual assistants.
Personal Use: As a versatile tool to assist with daily tasks, answer questions, and provide information.

Limitations

Potential Biases: Like all LLMs, Loxa Pro may reflect biases present in its training data.
Factual Accuracy: While highly accurate, the model may occasionally generate incorrect or misleading information. It is always recommended to verify information from multiple sources.
Resource Requirements: Although optimized, the model still requires a certain level of computational resources to run effectively.

How to Use

Example with transformers.pipeline:

from transformers import pipeline

messages = [
    {"role": "user", "content": "Write softmax formula in math style for me"},
]
pipe = pipeline("text-generation", model="explorewithai/LoxaPro", device_map = "cuda")
pipe(messages)

Downloads last month: 12

Safetensors

Model size

3B params

Tensor type

F32

Collection including frameai/LoxaPro

Loxa(Pro)

Collection

Loxa pro-vision is high quality models in Loxa family, they can run on small GPU. • 2 items • Updated Jan 19, 2025