Dromedary

Visit Tool

Dromedary is an open-source language model that focuses on principle-driven self-alignment. It provides a foundation for developing helpful, ethical, and reliable LLMs with minimal human supervision.

Claim this tool

No Views Yet

At a glance

Pricing

Open Source

Free tier

Yes

API

Skill level

Technical

About

What is Dromedary?

Dromedary is an open-source, self-aligned language model developed by IBM, designed to create helpful, ethical, and reliable Large Language Models (LLMs) with minimal human supervision. It features a principle-driven self-alignment process, including the updated Dromedary-2 (SFT) which utilizes diverse user prompts and an improved prompt structure for enhanced performance. Dromedary-2 also introduces the SALMON (Self-ALignMent with principle-fOllowiNg reward models) training pipeline for RLAIF. The project provides model weights as delta weights for LLaMA, synthetic data for self-alignment, and a full training pipeline for reproduction, making it a valuable resource for researchers and developers in the NLP domain.

Best used for

Ideal for developers and data scientists who need to build custom language models, conduct research into self-alignment techniques, and deploy ethical and reliable LLMs. Especially valuable for those seeking to minimize human supervision in the training process and leverage open-source resources.

Common actions

train language models

develop ethical AI

experiment with LLMs

deploy self-aligned models

face swapping"AI Agents"open-sourceautomated workflowgithub copilotcollaborationworkflowsdeepfakelow-code/no-code

Capabilities

Key features

Principle-driven self-alignment
Minimal human supervision
Dromedary-2 SFT training
SALMON RLAIF pipeline
LLaMA delta weights
Synthetic training data
Chatbot demo

Target Audience

developerdata scientistresearcher

Integrations

Not yet documented

Pricing & Plans

Open Source

Free

FAQs

What is the core principle behind Dromedary's self-alignment?

Dromedary's self-alignment is principle-driven, meaning it uses specific guidelines and an improved prompt structure to encourage the LLM to generate helpful, ethical, and reliable responses. This approach aims to reduce the need for extensive human supervision during training.

How does Dromedary-2 improve upon the original Dromedary model?

Dromedary-2 streamlines the self-alignment process to two stages, replacing the initial verbose cloning phase with diverse user prompts and an enhanced prompt with an additional exemplar. It also introduces the SALMON training pipeline for RLAIF, achieving better performance with less complexity.

Can Dromedary be used with LLaMA models?

Yes, Dromedary releases its weights as delta weights compatible with the LLaMA model license. Users can add these delta weights to original LLaMA weights to obtain Dromedary weights, allowing for deployment and inference with LLaMA base models.

Trending

Subcategories trending in Coding & Development

Code Assistants DevOps & Infrastructure No-Code / Low-Code Testing & QA Backend & APIs Prompt Engineering

Trending

Explore

Browse AI tools by category

Content & Design Productivity & Business Coding & Development AI Agents & Automation Research & Education Wellness & Lifestyle Career Development Marketing & Growth Data & Analytics Customer Support & CX Finance E-commerce