blexus 0.0.5

Blexus official package

🚀 Blexus

Welcome to Blexus, an AI innovation lab committed to crafting small, specialized AI models. Our mission is to unlock the full potential of AI through task-specific models that are fast, efficient, and highly customizable—ideal for developers, businesses, and everyday use cases. READ MORE

---

🛠️ Installation

You can install the Blexus package using pip. Run the following command:

pip install blexus

Use Quble models

from blexus import blexus
TextGenerator = blexus.TextGenerator

# Define the chat template  IMPORTANT: fill in the fields!
chat_template = {
    "before_system": "",
    "after_system": "",
    "before_user": "",
    "after_user": "",
    "before_assistant": ""
}

# Load Quble model
text_generator_quble = TextGenerator()
text_generator_quble.loadQuble()

# Generate text using Quble model
user_input = "Hello, how are you?"
system_input = "You are a helpful assistant."
generated_texts_quble = text_generator_quble.useQuble(user_input, system_input, chat_template, max_length=100, temperature=0.7, num_return_sequences=1)
print("Quble generated texts:", generated_texts_quble)

# Eject Quble model when done
text_generator_quble.eject_model()


# Load FCP model from a Hugging Face path
huggingface_path = "huggingface/fcp-model-name"  # Replace with actual Hugging Face model path
text_generator_fcp = TextGenerator()
text_generator_fcp.loadFCP(huggingface_path)

# Generate text using FCP model
generated_texts_fcp = text_generator_fcp.useFCP(user_input, system_input, chat_template, max_length=100, temperature=0.7, num_return_sequences=1)
print("FCP generated texts:", generated_texts_fcp)

# Eject FCP model when done
text_generator_fcp.eject_model()

Working exmaple:

from blexus import blexus
TextGenerator = blexus.TextGenerator

# Define the chat template  IMPORTANT: fill in the fields!
chat_template = {
    "before_system": "SYSTEM:",
    "after_system": "<|endofsystem|>",
    "before_user": "USER:",
    "after_user": "<|endoftext|>",
    "before_assistant": "ASSISTANT:"
}

# Load Quble model
text_generator_quble = TextGenerator()
text_generator_quble.loadQuble()

# Generate text using Quble model
user_input = "hi"
system_input = "You are a helpful assistant."
generated_texts_quble = text_generator_quble.useQuble(user_input, system_input, chat_template, max_length=50, temperature=0.7, num_return_sequences=1)
print("Quble generated texts:", generated_texts_quble)

# Eject Quble model when done
text_generator_quble.eject_model()

More coming soon!