AMD has released Instella-MoE-16B-A3B-Think, a new open-source Mixture-of-Experts language model. This model features 16 billion total parameters with 2.8 billion active parameters per token and was trained from scratch on AMD Instinct MI300X and MI325X GPUs. The release includes checkpoints for all training stages and provides instructions for integration with popular libraries like Transformers and vLLM, as well as tools such as SGLang and Docker. AI
IMPACT Provides a new open-source MoE model for researchers and developers, potentially driving innovation in model architectures and efficiency.
RANK_REASON New open-source model release from a major hardware vendor. [lever_c_demoted from frontier_release: ic=1 ai=1.0]
Read on Hugging Face Trending Models →
- AMD
- amd/Instella-MoE-16B-A3B-Think
- AMD Instinct MI300X
- AMD Instinct MI325X
- Docker
- Hugging Face
- SGLang
- Transformers
- vLLM
AI-generated summary · Google Gemini · from 1 sources. How we write summaries →
