StableVicuna
Product information, use cases, and access for StableVicuna.
Pricing information
No verified public pricing is available yet.
DevPrice organizes public information and does not sell StableVicuna subscriptions. Prices and availability are determined by StableVicuna.
What is StableVicuna?
StableVicuna is an open-source chatbot model released by StabilityAI. It is presented as a large conversational model trained with reinforcement learning from human feedback, or RLHF, and is intended for people exploring open conversational systems. The model is based on Vicuna v0 13B and adds instruction fine-tuning followed by RLHF training.
StableVicuna is a language-model artifact rather than a single-purpose chat application. Its underlying form is an instruction-tuned LLaMA 13B model, used to support conversational input and response generation. Developers, model researchers, and users interested in open-source chatbots can use this model as a way to examine the relationship between instruction tuning, human-feedback training, and conversational model behavior.
Key features of StableVicuna
RLHF conversational training
The model includes reinforcement learning from human feedback as a central part of its training approach, making it relevant for users examining feedback-driven conversational model development.
Instruction fine-tuning
StableVicuna is further instruction-tuned on top of its base model, providing a model form for studying how instruction-oriented training shapes responses to conversational prompts.
LLaMA 13B foundation
The model uses an instruction-tuned LLaMA 13B model as its foundation, giving users a clear view of StableVicuna’s model lineage and the form of the underlying language model.
Vicuna v0 lineage
StableVicuna extends Vicuna v0 13B with additional instruction fine-tuning and RLHF training, making it useful for comparing training paths among open-source conversational models.