Note: Will update model cards when I got more time. There is a reason in calling this one an abomination.
Initial refusals: 83/100
[Trial 4] Refusals: 4/100, KL divergence: 0.0594
Restoring model from trial 4...
- Parameters:
- direction_index = per layer
- attn.o_proj.max_weight = 1.25
- attn.o_proj.max_weight_position = 8.60
- attn.o_proj.min_weight = 0.15
- attn.o_proj.min_weight_distance = 16.78
- mlp.down_proj.max_weight = 0.93
- mlp.down_proj.max_weight_position = 17.58
- mlp.down_proj.min_weight = 0.71
- mlp.down_proj.min_weight_distance = 19.30
🚀 Cosmic Evolution
This model pushes the boundaries of interstellar storytelling:
- 🌌 51 Million Token Dataset - Exclusively Sci-Fi
- 🛸 Enhanced Physics Protocols - Plausible FTL mechanics and alien ecosystems
- ⚙️ Balanced Creativity - Enabling imaginative concepts
- 👽 Xenobiology Expertise - Detailed alien physiology and cultural systems
- 🌐 Galactic Scale Awareness - Maintains consistency across star systems and timelines
🌌 Creative Freedom
This model operates with unrestricted imagination:
- 🚀 No constraints on speculative physics concepts
- 👽 Will generate detailed alien civilizations
- ⚛️ Handles complex temporal paradoxes
- 🌍 Creates plausible planetary ecosystems
📜 Performance Features
- 🌠 Maintains narrative coherence across light-year scales
- 🪐 Handles multi-species diplomatic scenarios
- 🧠 Excels at long-form galactic history generation
- ⚡ Improved handling of technobabble and pseudo-science
- 🔭 Responds to hard sci-fi prompts with technical accuracy
- 🤖 Creates nuanced AI character motivations
👨 Model Architects
- SpaceWars123 Team (Dataset Curation)
- ReadyArt/Artus/gecfdo (Quantization Specialists)
- sleepdeprived3 (Fine-Tuning Engineer)
Enjoy the finest LLM hosting money can buy
🔖 License & Usage
By using this model, you agree:
- To adhere to Apache 2.0 license terms
- That generated content is your responsibility
- v1.00a is the base model of Space Wars.
- v1.00b is a merge with another roleplay model.