IkariDev/Athena-v2
IkariDev/Athena-v2 is a 13 billion parameter experimental language model developed by IkariDev and Undi95, built upon a complex merge of several 13B models including Xwin-LM, ReMM-v2.2, MLewd-L2, Slerpeno, and Magpie. This model is designed for general instruction-following tasks, utilizing an Alpaca prompt format. It offers a 4096-token context length and is suitable for a variety of conversational and text generation applications.
Loading preview...
IkariDev/Athena-v2: A Merged 13B Instruction-Following Model
IkariDev/Athena-v2 is an experimental 13 billion parameter language model, a collaborative effort between IkariDev and Undi95. This model is notable for its unique merging architecture, combining five distinct 13B base models to achieve its capabilities. The merge recipe integrates Xwin-LM/Xwin-LM-13B-V0.1, Undi95/ReMM-v2.2-L2-13B, Undi95/MLewd-L2-13B-v2-3, Brouz/Slerpeno, and boomerchan/Magpie-13b.
Key Capabilities
- Instruction Following: Designed to respond to instructions using the Alpaca prompt format.
- Merged Architecture: Leverages the strengths of multiple base models through a specific merging formula.
- Quantized Versions Available: Community-contributed quantized versions (GGUF, GPTQ, exl2, AWQ) are available for broader accessibility and deployment.
Good For
- General Text Generation: Suitable for various tasks requiring coherent and contextually relevant text output.
- Conversational AI: Its instruction-following nature makes it applicable for chatbot and interactive AI applications.
- Experimentation: Ideal for developers and researchers interested in exploring the performance of complex merged models.