Deathsquad10/TinyLlama-Remix

Deathsquad10/TinyLlama-Remix is a 1.1 billion parameter instruction-tuned causal language model, based on the TinyLlama architecture and Llama 2 tokenizer, with a 2048-token context length. It is fine-tuned following the Zephyr training recipe, initially on UltraChat and further aligned with DPO on the UltraFeedback dataset. This model is designed for chat applications, offering a compact solution for scenarios with restricted computational and memory resources.