Fine-tuning llama and qwen by modifying models' positional embeddings to enhance models' context length. - View it on GitHub