Large language models (LLMs) exhibit outstanding performance in machine translation via in-context learning. In contrast to sentence-level translation, document-level translation (DOCMT) by LLMs based on in-context learning faces two major challenges: firstly, document translations generated by LLMs are often incoherent; secondly, the length of demonstration for in-context learning is usually limited. To address these issues, we propose a Context-Aware Prompting method (CAP), which enables LLMs to generate more accurate, cohesive, and coherent translations via in-context learning. CAP takes into account multi-level attention, selects the most relevant sentences to the current one as context, and then generates a summary from these collected sentences. Subsequently, sentences most similar to the summary are retrieved from the datastore as demonstrations, which effectively guide LLMs in generating cohesive and coherent translations. We conduct extensive experiments across various DOCMT tasks, and the results demonstrate the effectiveness of our approach, particularly in zero pronoun translation (ZPT) and literary translation tasks.

Increasing the number of parameters in large language models (LLMs) usually improves performance in downstream tasks but raises compute and memory costs, making deployment difficult in resource-limited settings. Quantization techniques, which reduce the bits needed for model weights or activations with minimal performance loss, have become popular due to the rise of LLMs. However, most quantization studies use pre-trained LLMs, and the impact of quantization on instruction-tuned LLMs and the relationship between perplexity and benchmark performance of quantized LLMs are not well understood. Evaluation of quantized LLMs is often limited to language modeling and a few classification tasks, leaving their performance on other benchmarks unclear. To address these gaps, we propose a structured evaluation framework consisting of three critical dimensions: (1) knowledge & capacity, (2) alignment, and (3) efficiency, and conduct extensive experiments across ten diverse benchmarks. Our experimental results indicate that LLMs with 4-bit quantization can retain performance comparable to their non-quantized counterparts, and perplexity can serve as a proxy metric for quantized LLMs on most benchmarks. Furthermore, quantized LLMs with larger parameter scales can outperform smaller LLMs. Despite the memory savings achieved through quantization, it can also slow down the inference speed of LLMs. Consequently, substantial engineering efforts and hardware support are imperative to achieve a balanced optimization of decoding speed and memory consumption in the context of quantized LLMs.