Official pre-trained models and baselines in + MiniLLM: Knowledge distillation of LLMs during instruction tuning. + MiniPLM: Knowledge distillation of LLMs during pre-training.
Sizes Qwen's maker never released: distills, merges and other community models named after it.
