Although transformer architectures have achieved state-of-the-art performance across diverse domains, their quadratic computational complexity with respect to sequence length remains a significant bottleneck, particularly for latency-sensitive long-context applications. While recent linear-complexity alternatives are increasingly powerful, effectively training them from scratch is still resource-intensive. To overcome these limitations, we propose LAWCAT (Linear Attention with Convolution Across Time), a novel linearization framework designed to efficiently transfer the capabilities of pretrained transformers into a performant linear attention architecture. LAWCAT integrates causal Conv1D layers to enhance local dependency modeling and employs normalized gated linear attention to improve generalization across varying context lengths. Our comprehensive evaluations demonstrate that, distilling Mistral-7B with only 1K-length sequences yields over 90% passkey retrieval accuracy up to 22K tokens, significantly extending its effective context window. Similarly, Llama3.2-1B LAWCAT variant achieves competitive performance on S-NIAH 1&2&3 tasks (1K-8K context length) and BABILong benchmark (QA2&QA3, 0K-16K context length), requiring less than 0.1% pre-training tokens compared with pre-training models. Furthermore, LAWCAT exhibits faster prefill speeds than FlashAttention-2 for sequences exceeding 8K tokens. LAWCAT thus provides an efficient pathway to high-performance, long-context linear models suitable for edge deployment, reducing reliance on extensive long-sequence training data and computational resources.

We present a novel Parameter-Efficient Fine-Tuning (PEFT) method, dubbed as Adaptive Freezing of Low-Rank Adaptation (AFLoRA). Specifically, for each pre-trained frozen weight tensor, we add a parallel path of trainable low-rank matrices, namely a down-projection and an up-projection matrix, each of which is followed by a feature transformation vector. Based on a novel freezing score, we incrementally freeze these projection matrices during fine-tuning to reduce the computation and alleviate over-fitting. Our experimental results demonstrate that we can achieve state-of-the-art performance with an average improvement of up to 0.85% as evaluated on the GLUE benchmark while yielding up to 9.5× fewer average trainable parameters. While compared in terms of runtime, AFLoRA can yield up to 1.86× improvement as opposed to similar PEFT alternatives. Besides the practical utility of our approach, we provide insights on the trainability requirements of LoRA paths at different modules and the freezing schedule for the different projection matrices.