KaraKaraWitch/FindYourSwordInThisLand-Llama-3.3-72b
KaraKaraWitch/FindYourSwordInThisLand-Llama-3.3-72b is a 70 billion parameter language model merge, created by KaraKaraWitch using the SCE merge method. It is based on Llama-3.3-MagicalGirl-2 and integrates several other 70B models, including Dungeonmaster-V2.2-Expanded-LLaMa and Sapphira-L3.3. This model is an experimental merge, primarily focused on exploring the combination of various large language models for potentially nuanced conversational or creative applications, though its specific strengths are still under evaluation.
Loading preview...
Model Overview
KaraKaraWitch/FindYourSwordInThisLand-Llama-3.3-72b is a 70 billion parameter language model merge developed by KaraKaraWitch. This model was constructed using the SCE merge method with KaraKaraWitch/Llama-3.3-MagicalGirl-2 serving as the base model.
Merge Composition
The merge incorporates several other 70B parameter models, reflecting an experimental approach to combining different model characteristics. Key components include:
- Tarek07/Dungeonmaster-V2.2-Expanded-LLaMa-70B
- BruhzWater/Sapphira-L3.3-70b-0.1
- BruhzWater/Serpents-Tongue-L3.3-70b-0.3
- ReadyArt/L3.3-The-Omega-Directive-70B-Unslop-v2.0
- ReadyArt/Forgotten-Safeword-70B-v5.0
- Delta-Vector/Plesio-70B
- KaraKaraWitch/oiiaioiiai-B
Current Status and Intent
The creator notes that this merge is highly experimental, describing the writing as "sloppy" and indicating a potential restart from scratch. Performance benchmarks, such as UGI-Benchmark, are planned but not yet available, and the creator expresses doubt that it will significantly outperform its core components. This model is primarily an exploration of model merging techniques rather than a production-ready solution with defined capabilities.
Top 3 parameter combinations used by Featherless users for this model. Click a tab to see each config.