DarkArtsForge/Styx-12B
DarkArtsForge/Styx-12B is a 12 billion parameter MistralForCausalLM-based language model, developed by DarkArtsForge, with a 32768 token context length. This model is a prototype merge, created using the Della merge method from several 12B Heretic and MPOA models. It is characterized by its use of pre-ablated components, which may lead to certain refusals in its responses. Styx-12B is an experimental merge, serving as an earlier test before the development of Savage Sands.
Loading preview...
Styx-12B: A Prototype Merge
Styx-12B is a 12 billion parameter language model built on the MistralForCausalLM architecture, developed by DarkArtsForge. This model represents an early, experimental merge, predating the Savage Sands project, and was constructed using the Della merge method.
Key Characteristics
- Architecture: Based on
MistralForCausalLMwithmistralai--Mistral-Nemo-Instruct-2407as its base model. - Merge Method: Utilizes the
dellamerge method, combining components from multiple 12B models includingSorihon--Geodesic-Phantom-12B-Heretic,Sorihon--Nether-Moon-12B-Heretic,IIEleven11--Kalypso,EldritchLabs--Human-Like-Mistral-Nemo-Instruct-2407-MPOA,EldritchLabs--MN-12B-RP-Ink-Longform-MPOA,MuXodious-Rocinante-X-12B-v1-absolute-heresy, andPocketDoc--Dans-DangerousWinds-V1.1.0-12b. - Experimental Nature: As a prototype, it incorporates pre-ablated components, which may result in specific refusal behaviors during generation.
- Context Length: Supports a context window of 32768 tokens.
Intended Use
Styx-12B is primarily a technical artifact, useful for researchers and developers interested in exploring the outcomes of complex model merges, particularly those involving pre-ablated components. Its prototype status suggests it is best suited for experimental applications rather than production environments where consistent, refusal-free output is critical.