SteelStorage/Q2.5-MS-Mistoria-72b-v2
SteelStorage/Q2.5-MS-Mistoria-72b-v2 is a 72.7 billion parameter language model developed by SteelSkull, built upon the Qwen 2.5 architecture. This model is a merge of multiple base models, including Nexusflow/Athene-V2-Chat, EVA-UNIT-01/EVA-Qwen2.5-72B-v0.2, and shuttleai/shuttle-3. It is specifically designed to combine robust storytelling capabilities with maintained intelligence, making it suitable for applications requiring nuanced narrative generation. The model utilizes the Qwen format and supports a 32768 token context length.
Loading preview...
Q2.5-MS-Mistoria-72b-v2 Overview
SteelStorage/Q2.5-MS-Mistoria-72b-v2 is a 72.7 billion parameter language model developed by SteelSkull, based on the Qwen 2.5 architecture. This model represents SteelSkull's second iteration in developing a 72B model, with a core objective to integrate the strong storytelling abilities of various models while preserving overall intelligence.
Key Characteristics
- Architecture: Built on the Qwen 2.5 framework.
- Parameter Count: 72.7 billion parameters.
- Context Length: Supports a context window of 32768 tokens.
- Merging Strategy: Created using a "model_stock" merge method, combining:
Nexusflow/Athene-V2-ChatEVA-UNIT-01/EVA-Qwen2.5-72B-v0.2shuttleai/shuttle-3
- Format: Designed to be used with the Qwen format for prompts and responses.
Primary Goal
The model's development focuses on achieving a balance between sophisticated narrative generation and maintaining high levels of intelligence, making it suitable for tasks that require both creative and coherent output.
Quantization
GGUF quantizations are available from community members like bartowski and mradermacher, offering optimized versions for various deployment scenarios.