Bucket Pre-training is All You Need

Large language models (LLMs) have demonstrated exceptional performance across various natural language processing tasks. However, the conventional fixed-length data composition strategy for pretraining, which involves concatenating and splitting documents, can introduce noise and limit the model's ability to capture long-range dependencies. To address this, we first introduce three metrics for evaluating data composition quality: padding ratio, truncation ratio, and concatenation ratio. We further propose a multi-bucket data composition method that moves beyond the fixed-length paradigm, offering a more flexible and efficient approach to pretraining. Extensive experiments demonstrate that our proposed method could significantly improving both the efficiency and efficacy of LLMs pretraining. Our approach not only reduces noise and preserves context but also accelerates training, making it a promising solution for LLMs pretraining.

Review-LLM: Harnessing Large Language Models for Personalized Review Generation

Product review generation is an important task in recommender systems, which could provide explanation and persuasiveness for the recommendation. Recently, Large Language Models (LLMs, e.g., ChatGPT) have shown superior text modeling and generating ability, which could be applied in review generation. However, directly applying the LLMs for generating reviews might be troubled by the ``polite'' phenomenon of the LLMs and could not generate personalized reviews (e.g., negative reviews). In this paper, we propose Review-LLM that customizes LLMs for personalized review generation. Firstly, we construct the prompt input by aggregating user historical behaviors, which include corresponding item titles and reviews. This enables the LLMs to capture user interest features and review writing style. Secondly, we incorporate ratings as indicators of satisfaction into the prompt, which could further improve the model's understanding of user preferences and the sentiment tendency control of generated reviews. Finally, we feed the prompt text into LLMs, and use Supervised Fine-Tuning (SFT) to make the model generate personalized reviews for the given user and target item. Experimental results on the real-world dataset show that our fine-tuned model could achieve better review generation performance than existing close-source LLMs.

xCoT: Cross-lingual Instruction Tuning for Cross-lingual Chain-of-Thought Reasoning

Chain-of-thought (CoT) has emerged as a powerful technique to elicit reasoning in large language models and improve a variety of downstream tasks. CoT mainly demonstrates excellent performance in English, but its usage in low-resource languages is constrained due to poor language generalization. To bridge the gap among different languages, we propose a cross-lingual instruction fine-tuning framework (xCOT) to transfer knowledge from high-resource languages to low-resource languages. Specifically, the multilingual instruction training data (xCOT-INSTRUCT) is created to encourage the semantic alignment of multiple languages. We introduce cross-lingual in-context few-shot learning (xICL)) to accelerate multilingual agreement in instruction tuning, where some fragments of source languages in examples are randomly substituted by their counterpart translations of target languages. During multilingual instruction tuning, we adopt the randomly online CoT strategy to enhance the multilingual reasoning ability of the large language model by first translating the query to another language and then answering in English. To further facilitate the language transfer, we leverage the high-resource CoT to supervise the training of low-resource languages with cross-lingual distillation. Experimental results on previous benchmarks demonstrate the superior performance of xCoT in reducing the gap among different languages, highlighting its potential to reduce the cross-lingual gap.

PEPT: Expert Finding Meets Personalized Pre-training

Finding appropriate experts is essential in Community Question Answering (CQA) platforms as it enables the effective routing of questions to potential users who can provide relevant answers. The key is to personalized learning expert representations based on their historical answered questions, and accurately matching them with target questions. There have been some preliminary works exploring the usability of PLMs in expert finding, such as pre-training expert or question representations. However, these models usually learn pure text representations of experts from histories, disregarding personalized and fine-grained expert modeling. For alleviating this, we present a personalized pre-training and fine-tuning paradigm, which could effectively learn expert interest and expertise simultaneously. Specifically, in our pre-training framework, we integrate historical answered questions of one expert with one target question, and regard it as a candidate aware expert-level input unit. Then, we fuse expert IDs into the pre-training for guiding the model to model personalized expert representations, which can help capture the unique characteristics and expertise of each individual expert. Additionally, in our pre-training task, we design: 1) a question-level masked language model task to learn the relatedness between histories, enabling the modeling of question-level expert interest; 2) a vote-oriented task to capture question-level expert expertise by predicting the vote score the expert would receive. Through our pre-training framework and tasks, our approach could holistically learn expert representations including interests and expertise. Our method has been extensively evaluated on six real-world CQA datasets, and the experimental results consistently demonstrate the superiority of our approach over competitive baseline methods.