lmsys/vicuna-13b-delta-v0

TEXT GENERATIONConcurrency Cost:1Model Size:13BQuant:FP8Ctx Length:4kPublished:Apr 3, 2023Architecture:Transformer0.5K Cold

The lmsys/vicuna-13b-delta-v0 is a 13 billion parameter delta model developed by LMSYS, fine-tuned from the LLaMA architecture. This model is specifically designed as a chat assistant, trained on user-shared conversations from ShareGPT. It serves primarily as a research tool for large language models and chatbots, requiring application on original LLaMA weights to become functional.

Loading preview...

Vicuna-13B-Delta-v0 Overview

Vicuna-13B-Delta-v0 is a 13 billion parameter language model developed by LMSYS, fine-tuned from the LLaMA architecture. This specific release is a "delta model," meaning it cannot be used directly but must be applied on top of the original LLaMA weights to form the complete Vicuna model. It is an auto-regressive language model based on the transformer architecture, primarily intended for research purposes in large language models and chatbots.

Key Capabilities

  • Chat Assistant: Fine-tuned to function as a conversational AI, leveraging user-shared dialogues from ShareGPT.com.
  • Research Tool: Designed for researchers and hobbyists exploring natural language processing, machine learning, and artificial intelligence.
  • Instruction Fine-tuning: Utilizes supervised instruction fine-tuning on approximately 70,000 conversations.

Good For

  • Academic Research: Ideal for studying the behavior and performance of instruction-tuned large language models.
  • Chatbot Development: Provides a base for developing and experimenting with conversational AI applications.
  • LLaMA-based Projects: Suitable for users who already possess LLaMA weights and wish to build upon them with Vicuna's chat capabilities.