If you’re looking to enhance your storytelling capabilities through the use of AI, then Waxwing is your go-to tool! This blog will guide you through using the Waxwing Storytelling model, developed for the Llama 2 70B version. With its unique features, you can create engaging narratives with ease. Whether you’re a seasoned writer or a neophyte, this guide has got you covered!
What is Waxwing?
Waxwing is a storytelling model that incorporates a turn-based instruction system, allowing users to guide their stories seamlessly. Tailor the feel of your narrative using style tags, resulting in authentic storytelling without common AI biases like positivity bias or rushed conclusions. This ensures that your story flows naturally, resembling human-generated content.
Getting Started with Waxwing
To begin utilizing the Waxwing model, here’s a breakdown of the necessary steps:
- Access the Model: Waxwing is a LoRA (Low-Rank Adaptation) model that you can apply at runtime on any variant of the Llama 2 70B base model. You can also find the 16fp merged model in full precision here.
- Quantized Models: For enhanced performance with Exllama 2, choose one of the available quantized models such as 2.5bpw, 3.0bpw, 4.65bpw, 6.0bpw, and 8.0bpw. Explore these options here.
Using Style Tags
One of the standout features of Waxwing is the use of style tags, which allows for customization of the writing style. Here’s how to implement it:
- Begin your writing instruction with
Write a scene where: - Incorporate style tags, which should follow a specific format. An example of the format would be:
Tone: tense, mysterious. Writing style: dramatic. Written in abstract detail. - For extended writing, simply type
continueto carry on the narrative.
Training the Model
The Waxwing model underwent a rigorous training regimen on a machine equipped with 72GB of VRAM, ensuring robust performance. If you’re curious about the training parameters, they are available in the training_parameters.json file of the main branch. The model was developed using FartyPants Training_PRO extension for the Oobabooga Text Generation WebUI.
Troubleshooting Ideas
If you encounter issues while working with Waxwing, here are some troubleshooting tips:
- Model Not Responding: Ensure you have the correct model version applied. Check for any missing dependencies that may affect performance.
- Style Tags Not Working: Make sure you are following the correct formatting for style tags. Experiment with different combinations to see how they affect the narrative.
- Output Quality: If the output does not meet your expectations, remember to adjust your input prompts for clarity and detail. The more context you provide, the better the response will be.
- For more insights, updates, or to collaborate on AI development projects, stay connected with fxis.ai.
Conclusion
In conclusion, Waxwing is a powerful tool for anyone interested in storytelling, enabling personalized narrative crafting without the limitations of traditional AI models. By following the steps outlined above and utilizing the distinct features of Waxwing, you can start writing stories that captivate and engage.
At fxis.ai, we believe that such advancements are crucial for the future of AI, as they enable more comprehensive and effective solutions. Our team is continually exploring new methodologies to push the envelope in artificial intelligence, ensuring that our clients benefit from the latest technological innovations.

