DS-Archive/lzlv-limarpv3-l2-70b
DS-Archive/lzlv-limarpv3-l2-70b is a 69 billion parameter Llama 2-based model, created by DS-Archive, resulting from a merge between lizpreciatior/lzlv_70b_fp16_hf and a PEFT adapter trained on the LimaRP dataset. With a 32768 token context length, this model is specifically fine-tuned for advanced roleplaying chat scenarios, supporting detailed character personas and user interactions. Its primary differentiator is the inclusion of a message length control feature, allowing users to specify response lengths from 'micro' to 'unlimited' for tailored roleplay experiences.
Loading preview...
Model Overview
DS-Archive/lzlv-limarpv3-l2-70b is a 69 billion parameter language model built upon the Llama 2 architecture. It is a merge of lizpreciatior/lzlv_70b_fp16_hf and a PEFT adapter derived from the LimaRP dataset, specifically Doctor-Shotgun/limarpv3-llama2-70b-qlora, with a 0.5 weight. This model is designed with a substantial context length of 32768 tokens.
Key Capabilities
- Advanced Roleplaying: Optimized for engaging in detailed roleplaying chats, allowing for rich character and user persona definitions within scenarios.
- Alpaca Instruction Format: Utilizes the Alpaca instruction format, specifically adapted from LimaRP v3, for structured prompts.
- Response Length Control: A unique feature enabling users to specify the desired length of bot responses using modifiers like
micro,medium,long, orunlimited. This directly influences the verbosity of the AI's output.
Intended Use Cases
This model is primarily intended for niche roleplaying applications. It is important to note that due to its training and fine-tuning, it is not suitable for providing factual information or advice. Users should be aware of potential biases inherited from both the base model and the roleplaying forum data it was trained on.