WizardLM-13b-v1.2

WizardLM-13B-V1.2 is a large language model developed by the WizardLM Team that follows complex instructions and provides detailed, helpful responses. It is part of the broader WizardLM project, which includes various models tailored for different tasks such as coding and mathematical problem-solving.

Key features:
- Follows complex instructions, making it suitable for tasks that require detailed and nuanced responses
- Supports multi-turn conversations, adopting the prompt format from Vicuna for more interactive and context-aware dialogues
- Achieves a score of 7.06 on the MT-Bench Leaderboard, 89.17% on the AlpacaEval Leaderboard, and 101.4% on the WizardLM Eval, indicating high performance in various evaluation settings
- Generates coherent and contextually relevant text based on the input it receives

How it works:
Users interact with WizardLM-13B-V1.2 primarily through text prompts. The model interprets and responds to user inputs in a conversational manner, providing detailed and helpful responses while adhering to the format of a polite and informative assistant.

Integrations:
WizardLM-13B-V1.2 can be integrated into various applications and platforms through its API. It is available on Hugging Face, providing easy access to the model for developers and researchers.

Use of AI:
WizardLM-13B-V1.2 leverages generative artificial intelligence to produce text based on user inputs.

AI foundation model:
WizardLM-13B-V1.2 is built on the Llama 2 foundation model, which provides a robust base for its advanced capabilities. The model's training involves extensive datasets and fine-tuning to ensure high performance in following complex instructions and generating relevant responses.

How to access:
- The model is available as a web app, API, and can be accessed through platforms like Hugging Face.
- It is suitable for developers, researchers, and businesses that require advanced language models for tasks such as coding, content generation, and customer support.
- The WizardLM-13B-V1.2 model was released on July 25, 2023.

The model weights for WizardLM-13B-V1.2 are open-source, but the data used for training is under stricter auditing and review, and not publicly released without authorization.