Wizardlm Uncensored 13B - Details

Wizardlm Uncensored 13B is a large language model developed by Cognitive Computations, a community-driven initiative. It features 13 billion parameters, making it a robust tool for complex tasks. The model operates under the Llama 2 Community License Agreement (LLAMA-2-CLA), ensuring accessibility while adhering to specific usage guidelines. Designed as an uncensored version of Llama 2, it emphasizes reduced biases and broader applicability across diverse scenarios.

Wizardlm Uncensored 13B is a variant of the WizardLM series trained on a dataset subset where alignment or moralizing responses were removed, aiming to create an uncensored model without built-in ethical constraints. This design allows for custom alignment through methods like RLHF LoRA, enabling users to add specific guidelines post-training. The model operates without guardrails, meaning it lacks safety mechanisms to filter outputs. Users are fully responsible for any content generated or published, akin to accountability for actions involving dangerous tools like knives or guns. The model’s outputs reflect the user’s input and choices, with no liability shifted to the system itself.

Wizardlm Uncensored 13B has 13 billion parameters, placing it in the mid-scale category of open-source LLMs, offering a balance between performance and resource efficiency for moderate complexity tasks. Its 4k token context length falls into the short context range, making it suitable for concise interactions but limiting its ability to handle extended or highly detailed texts. The model’s design prioritizes accessibility and flexibility, allowing users to tailor alignment and guardrails post-training, though this also means it lacks built-in safety mechanisms. The 13b parameter size ensures it can manage a wide range of tasks without excessive computational demands, while the 4k context length restricts its effectiveness for very long documents or continuous dialogue.

• Parameter Size: 13b
• Context Length: 4k

Wizardlm Uncensored 13B is a model that could be used for a range of tasks, though its potential applications require careful exploration. Its 13b parameter size and 4k context length make it a possible tool for research, where users might test hypotheses or analyze data without predefined constraints. It could also serve as a possible resource for content creation, allowing users to generate text for creative projects or experimentation. In educational settings, it might be a possible aid for students or educators seeking to explore language patterns or develop interactive learning materials. However, these uses are possible and not guaranteed, as the model’s uncensored nature means outputs depend entirely on user input and context. The lack of built-in guardrails means any application must be thoroughly investigated to ensure alignment with ethical and practical goals.

• research
• content creation
• educational purposes

Wizardlm Uncensored 13B could be a possible tool for tasks requiring flexibility and adaptability, though its applications are still under exploration. It might be a possible resource for creative writing or content generation, where users seek to experiment with unfiltered outputs. It could also be a possible aid for academic research, allowing for hypothesis testing or data analysis without predefined constraints. Additionally, it might be a possible platform for educational experiments, enabling interactive learning scenarios or language exploration. These uses are possible and not guaranteed, as the model’s uncensored nature means outputs depend entirely on user input and context. Each application must be thoroughly evaluated and tested before deployment to ensure alignment with specific goals.

• research
• content creation
• educational purposes

Wizardlm Uncensored 13B with the medium q4 quantization requires a GPU with at least 16GB-32GB VRAM to run efficiently, depending on the workload and system configuration. This version balances precision and performance, making it suitable for users with mid-range hardware. The 13b parameter size and q4 quantization reduce memory demands compared to higher-precision variants, but still necessitate a capable GPU for smooth operation. Always verify your system’s VRAM and cooling capabilities before deployment.

• fp16, q2, q3, q4, q5, q6, q8

Wizardlm Uncensored 13B is a large language model with 13 billion parameters trained to reduce biases by removing alignment-focused content, operating under the Llama 2 Community License Agreement (LLAMA-2-CLA). Its uncensored nature means it lacks built-in guardrails, requiring users to handle responsibility for outputs and implement custom alignment mechanisms like RLHF LoRA.

Huggingface Model Page
Ollama Model Page