sophosympatheia/Mero-Artemis-31B-v0.3.1
Mero-Artemis-31B-v0.3.1 is an experimental 31 billion parameter language model created by sophosympatheia, resulting from a merge of BeaverAI's Artemis-31B-v1h and zerofata's G4-MeroMero-31B. This model aims to combine the distinct personality and style of Artemis with MeroMero's strong attention to detail. It features a 32768 token context length and is designed for nuanced conversational and detailed text generation tasks.
Loading preview...
Model Overview
Mero-Artemis-31B-v0.3.1 is an experimental 31 billion parameter language model developed by sophosympatheia. This model is a merge of two distinct large language models: BeaverAI's Artemis-31B-v1h and zerofata's G4-MeroMero-31B. The merge process involved using a Q8_0 GGUF file for Artemis due to the unavailability of full precision FP16 weights, a method chosen to balance precision and merge feasibility.
Key Characteristics
- Hybrid Personality: The primary goal of this merge was to combine the unique personality and stylistic traits of the Artemis model with the strong attention to detail characteristic of the MeroMero model.
- Experimental Nature: This release is explicitly labeled as an experiment, with the developer seeking community feedback before a potential formal release.
- Context Length: The model supports a substantial context window of 32768 tokens, allowing for processing and generating longer, more complex texts.
Recommended Sampler Settings
For initial testing, the developer suggests the following sampler settings:
- Temperature: 1.0
- Min-P: 0.05 - 0.1
- DRY: 0.8 multiplier, 1.8 base
- Adaptive-P: 0.6 target, 0.5 decay