Phr00t/Phr00tyMix-v4-32B

Phr00tyMix-v4-32B Overview

Phr00t/Phr00tyMix-v4-32B is a 32.8 billion parameter language model developed by Phr00t, created through a sophisticated merge of several pre-trained models using the Model Stock method. This version builds upon its predecessor, Phr00tyMix-v3-32B, with a specific focus on enhancing instruction following and overall coherency, which were areas identified for improvement in the previous iteration.

Key Capabilities

• Improved Instruction Following: Designed to better adhere to user instructions and prompts, leading to more predictable and controlled outputs.
• Enhanced Coherency: Aims to produce more logically structured and consistent narratives, improving the overall quality of generated text.
• Maintained Creativity: Despite the focus on coherency, the model retains the high creativity observed in earlier Phr00tyMix versions.
• Uncensored Output: Capable of generating uncensored content when explicitly requested, offering flexibility for various applications.

Merge Details

This model was constructed by merging several distinct models, including allura-org/Qwen2.5-32b-RP-Ink, nicoboss/DeepSeek-R1-Distill-Qwen-32B-Uncensored, Delta-Vector/Archaeo-32B-KTO, arcee-ai/Virtuoso-Medium-v2, and Phr00t/Phr00tyMix-v2-32B, with Phr00t/Phr00tyMix-v3-32B serving as the base model. This multi-model integration contributes to its balanced performance across creativity and instruction adherence.

Good For

• Storytelling and Creative Writing: Its blend of creativity and improved coherency makes it well-suited for generating engaging and structured narratives.
• Roleplay and Interactive Applications: The enhanced instruction following and obedience are beneficial for applications requiring responsive and context-aware interactions.
• Applications Requiring Uncensored Content: Suitable for use cases where the ability to generate uncensored text is a specific requirement.