Model Overview

athirdpath/NSFW_DPO_Noromaid-7b is a 7 billion parameter language model, building upon the NeverSleep/Noromaid-7b-v0.1.1 base. This iteration has undergone further fine-tuning using Direct Preference Optimization (DPO) on a specific NSFW dataset, following an initial training phase on a DPO-v2 dataset. The model maintains an 8192-token context window.

Key Capabilities

• Specialized Content Generation: Primarily designed and optimized for generating explicit or adult-oriented text content.
• DPO Fine-tuning: Leverages Direct Preference Optimization for aligning outputs with specific preferences, particularly for NSFW contexts.
• Noromaid Architecture: Based on the Noromaid model family, providing a foundation for its language generation capabilities.

Good For

• Explicit Text Generation: Ideal for use cases that specifically require the creation of NSFW or adult-themed textual content.
• Research into DPO for Niche Content: Can be used by researchers exploring the effectiveness of DPO on highly specialized and sensitive datasets.