Dromedary
IBM Dromedary: An open-source, self-aligned LLM for helpful, ethical, and reliable AI.
Quick facts
- Best for
- IBM Dromedary: An open-source, self-aligned LLM for helpful, ethical, and reliable AI.
- Pricing
- Freemium
- Editor rating
- 4.5 / 5
- Community saves
- 0
About Dromedary
IBM Dromedary is an open-source, self-aligned large language model trained on LLaMA and LLaMA‑2 bases using principle-driven SELF-ALIGN methods (SFT and RLAIF via SALMON) to be helpful, ethical, and reliable with minimal human supervision. The project releases delta/QLoRA weights and synthetic datasets on Hugging Face, provides full training and inference pipelines (including multi-GPU model-parallel deployment), and is documented alongside a NeurIPS 2023 Spotlight paper.
Pros
- Open-source self-aligned LLM from IBM’s Self-Align team
- Principle-driven alignment with minimal human supervision
- SELF-ALIGN pipeline with Topic-Guided Red-Teaming Self-Instruct
- Dromedary-2 SFT: simplified two-stage process with FastChat exemplar
- Dromedary-2 RLAIF: SALMON reward-model pipeline on LLaMA-2-70B
- Delta weights (LoRA/QLoRA) compatible with LLaMA and LLaMA-2
- Synthetic datasets released on Hugging Face (65B and 70B SFT)
- Custom llama_dromedary package for training and inference
- Model-parallel, multi-GPU inference faster than HF pipeline-parallel
- End-to-end training and inference guides with chatbot demo
- Support for 1, 2, 4, 8, or 16 GPU deployments using original LLaMA repo
- GPL-3.0 licensed repository with NeurIPS 2023 Spotlight paper reference