L3.3-70B-Animus-V10.0-EXL2

Send me your support to help me feed the data beast! also taking comissions for universe specific models

Support on Ko-fi

Quantized Models

The quantized model files are available for download. Click the buttons below to view the files.

Download GGUF Files → Download EXL2 Files → Download EXL3 Files →

Character Card & Lore Book

For the best roleplaying experience, it is highly recommended to use the provided character card and lore book. These files help guide the model's persona and provide rich, in-universe context.

Download Files →

SillyTavern Sampler Presets

For a seamless setup in SillyTavern, you can download pre-configured sampler presets. These are tuned to provide an optimal balance between creativity and narrative coherence for this model.

Simply download the .json file below and import it into SillyTavern's sampler presets menu.

Download SillyTavern Presets →

Model Description

This is Version 10.0, a new pinnacle in the Animus series. Rather than a simple incremental update, V10.0 is a synergistic merge of several previous versions, designed to combine the best traits of each into a single, highly capable model.

The result is a surprisingly robust creative partner that reviewers have called the best version yet. It was created by merging the following models:

• L3.3-70B-Animus-Base: The foundational model, providing core intelligence and a broad understanding of the Wings of Fire universe.
• Version 6.1 & 7.0: Widely regarded as the best previous versions for lore accuracy and authentic character portrayals.
• Version C & D: Two experimental fine-tunes that successfully introduced higher stakes and consequences, including the possibility of user character death, which was missing from V7.0.

By blending these components, V10.0 achieves the narrative depth of V7.0 while incorporating the dramatic potential of the C/D series. A surprising outcome is that this specialized blend has made the model exceptionally capable at general, non-WOF roleplay, making it the most versatile Animus model to date.

Training Details

A Merged Model: The V10.0 Lineage

V10.0 was created through a model merging process rather than a traditional fine-tune. This approach was taken because different training runs produced models with unique, desirable strengths that were difficult to replicate in a single training session.

The lineage is as follows:

1. The Darkhn/L3.3-70B-Animus-Base model was created by first training on general completion data on a first stage, then for stage 2, fine-tuning on Wings of Fire roleplay data, Which gave
2. C and D to specifically add these missing elements, but felt like they were "missing a little spark" on their own.
3. V7.0 became the benchmark for quality, but lacked certain narrative possibilities (like user character death).
4. V10.0 is the final product, a stock merge of V6.1, V7.0, C, and D, balancing the strengths of all contributors.

Feature Update: Removal of DM Choices

A key feature in previous test versions—the presentation of multiple-choice actions (e.g., A, B, C) to guide the user—has been removed from V10.0.

While a promising concept, this feature needs further refinement to ensure it enhances, rather than restricts, the roleplaying experience. It may be reintroduced in a more polished form in a future release. For now, the model returns to a more traditional, open-ended prose format.

Training Data (for Component Models)

The component models (V7.0, C, D) were fine-tuned on a high-quality dataset of 3,200 examples with several key characteristics:

• Canon-Centric Scenarios: All roleplay scenarios are based on pivotal events from the Wings of Fire book series, exploring "what-if" outcomes. (e.g., What if Darkstalker didn't kill Arctic at that moment?).
• Canon-Only Characters: The models were trained exclusively on canon characters from the books. AI-generated characters were removed from the training data (except for the user's persona).
• Improved Data Cleaning: The dataset underwent a rigorous cleaning process to remove formatting artifacts from previous versions, such as **scene transitions**, resulting in a cleaner and more natural narrative style.
• Refined Turn Structure: Addressed an issue where consecutive AI turns appeared in the dataset, leading to a healthier learning curve and more natural conversational flow.

Intended Use & Limitations

• Intended Use: The primary purpose of this model is for creative and roleplaying within the Wings of Fire universe. However, user feedback indicates it is also highly effective for general-purpose roleplaying.
• Limitations & Quirks:
  • Performance on tasks outside of its training domain (general knowledge, coding, etc.) is not guaranteed and will likely be poor.
  • Versatility: While it appears to be only a Wings of Fire tuned model, users have reported it is very capable of performing normal roleplay with other settings and characters.
  • The model may "hallucinate" or generate plausible but non-canonical information, especially when pushed outside the established "what-if" scenarios.
  • Content: The training data includes mature and darker themes from the Wings of Fire series, such as conflict, character death, and moral ambiguity. The model is capable of generating content reflecting these themes. As always, it is up to the user what they do with it.
  • Formatting: Training data was cleaned to remove narrative artifacts like **scene transitions**. The model should now produce cleaner prose.
  • Safety: This model has not undergone additional safety alignment beyond what was included in its base Llama 3.3 model. Standard responsible AI practices should be followed.

Acknowledgements

• Credit to Meta for the powerful Llama 3.3 architecture.
• Credit to Google for the Gemini Pro model, used in dataset generation.
• Credit to Evan Armstrong for Augmentoolkit, an invaluable tool for dataset creation.