In this work, we introduce ChatQA 2, an Llama 3.0-based model with a 128K context window, designed to bridge the gap between open-source LLMs and leading proprietary models (e.g., GPT-4-Turbo-2024-04-09) in long context un- derstanding and retrieval-augmented generation (RAG) capabilities. These two capabilities are complementary to each other and essential for LLMs to process large volumes of information that cannot fit into a single prompt. We present a detailed continued training recipe to extend the context window of Llama3- 70B-base from 8K to 128K tokens, along with a three-stage instruction tun- ing process to enhance the model’s instruction-following, RAG performance, and long-context understanding capabilities. Our results demonstrate that the Llama3-ChatQA-2-70B model outperforms most existing state-of-the-art models, including GPT-4-Turbo-2024-04-09, Qwen2-72B-Instruct, and Llama3.1-70B- Instruct, on ultra-long tasks beyond 100K tokens, as well as on the RAG benchmark using only a 4K context window, showing the strong long context capability across varying sequence lengths. We further provide extensive comparisons between direct long-context and RAG solutions using the same state-of-the-art long-context LLMs. Interestingly, we find that the performance of strong long-context LLMs using RAG improves when retrieving a larger number of chunks. With a large set of top-k chunks, RAG consistently outperforms direct long-context solution using the same state-of-the-art long-context models (e.g., Llama3-ChatQA-2-70B and Qwen2-72B-Instruct) on both 32K and 128K benchmarks. We open-source the model weights, training data, and the evaluation setup for the for the community: https://chatqa2-project.github.io/

Despite recent progress in large-scale reinforcement learning (RL) for reasoning, the training recipe for building high-performing reasoning models remains elusive. Key implementation details of frontier models, such as DeepSeek-R1, including data curation strategies and RL training recipe, are often omitted. Moreover, recent research indicates distillation remains more effective than RL for smaller models. In this work, we demonstrate that large-scale RL can significantly enhance the reasoning capabilities of strong, small- and mid-sized models, achieving results that surpass those of state-of-the-art distillation-based models. We systematically study the RL training process through extensive ablations and propose a simple yet effective approach: first training on math-only prompts, then on code-only prompts. Notably, we find that math-only RL not only significantly enhances the performance of strong distilled models on math benchmarks (e.g., +14.6\% / +17.2\% on AIME 2025 for the 7B / 14B models), but also code reasoning tasks (e.g., +6.8\% / +5.8\% on LiveCodeBench for the 7B / 14B models). In addition, extended code-only RL further improves code benchmark performance while causing minimal degradation in math results. We develop a robust data curation pipeline to collect challenging prompts with high-quality, verifiable answers and test cases to enable verification-based RL across both domains. The dataset will be released to support open research. Finally, we identify key experimental insights, including curriculum learning with progressively increasing response lengths and the stabilizing effect of on-policy parameter updates. We find that RL not only elicits the foundational reasoning capabilities acquired during pretraining and supervised fine-tuning (SFT), but also pushes the limits of the model’s reasoning ability, enabling it to solve problems that were previously unsolvable.

Extending the context window of large language models (LLMs) is getting popular recently, while the solution of augmenting LLMs with retrieval has existed for years. The natural questions are: i) Retrieval-augmentation versus long context window, which one is better for downstream tasks? ii) Can both methods be combined to get the best of both worlds? In this work, we answer these questions by studying both solutions using two state-of-the-art pretrained LLMs, i.e., a proprietary 43B GPT and Llama2-70B. Perhaps surprisingly, we find that LLM with 4K context window using simple retrieval-augmentation at generation can achieve comparable performance to finetuned LLM with 16K context window via positional interpolation on long context tasks, while taking much less computation. More importantly, we demonstrate that retrieval can significantly improve the performance of LLMs regardless of their extended context window sizes. Our best model, retrieval-augmented Llama2-70B with 32K context window, outperforms GPT-3.5-turbo-16k and Davinci003 in terms of average score on nine long context tasks including question answering, query-based summarization, and in-context few-shot learning tasks. It also outperforms its non-retrieval Llama2-70B-32k baseline by a margin, while being much faster at generation. Our study provides general insights on the choice of retrieval-augmentation versus long context extension of LLM for practitioners.

In this work, we introduce ChatQA, a suite of models that outperform GPT-4 on retrieval-augmented generation (RAG) and conversational question answering (QA). To enhance generation, we propose a two-stage instruction tuning method that significantly boosts the performance of RAG. For effective retrieval, we introduce a dense retriever optimized for conversational QA, which yields results comparable to the alternative state-of-the-art query rewriting models, while substantially reducing deployment costs. We also present the ChatRAG Bench, which encompasses ten datasets covering comprehensive evaluations on RAG, table-related QA, arithmetic calculations, and scenarios involving unanswerable questions. Our ChatQA-1.0-70B (score: 54.14), built on Llama2, a weaker foundation model than GPT-4, can slightly outperform GPT-4-0613 (score: 53.90) and GPT-4-Turbo-2024-04-09 (score: 54.03) on the ChatRAG Bench, without relying on any synthetic data from OpenAI GPT models. Notably, Llama3-ChatQA-1.5-70B model surpasses the accuracy of GPT-4-Turbo-2024-04-09 by a margin. These results demonstrate the exceptional quality of the proposed ChatQA recipe. To advance research in this field, we open-sourced the model weights, instruction tuning data, ChatRAG Bench, and retriever for the community.

Large language models (LLMs) typically utilize the top-k contexts from a retriever in retrieval-augmented generation (RAG). In this work, we propose a novel method called RankRAG, which instruction-tunes a single LLM for both context ranking and answer generation in RAG. In particular, the instruction-tuned LLMs work surprisingly well by adding a small fraction of ranking data into the training blend, and outperform existing expert ranking models, including the same LLM exclusively fine-tuned on a large amount of ranking data. For generation, we compare our model with many strong baselines, including ChatQA-1.5, an open-sourced model with the state-of-the-art performance on RAG benchmarks. Specifically, our Llama3-RankRAG-8B and Llama3-RankRAG-70B significantly outperform Llama3-ChatQA-1.5-8B and Llama3-ChatQA-1.5-70B, respectively, on nine general knowledge-intensive benchmarks for RAG. In addition, it also performs comparably to GPT-4 on five RAG benchmarks in the biomedical domain without instruction fine-tuning on biomedical data, demonstrating its superb capability for generalization to new domains.

Large-scale vision-language pre-trained (VLP) models are prone to hallucinate non-existent visual objects when generating text based on visual information.In this paper, we systematically study the object hallucination problem from three aspects.First, we examine recent state-of-the-art VLP models, showing that they still hallucinate frequently, and models achieving better scores on standard metrics (e.g., CIDEr) could be more unfaithful.Second, we investigate how different types of image encoding in VLP influence hallucination, including region-based, grid-based, and patch-based.Surprisingly, we find that patch-based features perform the best and smaller patch resolution yields a non-trivial reduction in object hallucination.Third, we decouple various VLP objectives and demonstrate that token-level imagetext alignment and controlled generation are crucial to reducing hallucination.Based on that, we propose a simple yet effective VLP loss named ObjMLM to further mitigate object hallucination.Results show that it reduces object hallucination by up to 17.4% when tested on two benchmarks (COCO Caption for in-domain and NoCaps for out-of-domain evaluation).

Boxin Wang, Wei Ping, Peng Xu, Lawrence McAfee, Zihan Liu, Mohammad Shoeybi, Yi Dong, Oleksii Kuchaiev, Bo Li, Chaowei Xiao, Anima Anandkumar, Bryan Catanzaro. Proceedings of the 2023 Conference on Empirical Methods in Natural Language Processing. 2023.

Speech emotion recognition plays a crucial role in human-computer interactions. However, most speech emotion recognition research is biased toward English-speaking adults, which hinders its applicability to other demographic groups in different languages and age groups. In this work, we analyze the transferability of emotion recognition across three different languages-English, Mandarin Chinese, and Cantonese; and 2 different age groups-adults and the elderly. To conduct the experiment, we develop an English-Mandarin speech emotion benchmark for adults and the elderly, BiMotion, and a Cantonese speech emotion dataset, YueMotion. This study concludes that different language and age groups require specific speech features, thus making cross-lingual inference an unsuitable method. However, cross-group data augmentation is still beneficial to regularize the model, with linguistic distance being a significant influence on cross-lingual transferability. We release publicly release our code at https://github.com/HLTCHKUST/elderly_ser.

Code-switching is a speech phenomenon occurring when a speaker switches language during a conversation. Despite the spontaneous nature of code-switching in conversational spoken language, most existing works collect code-switching data from read speech instead of spontaneous speech. ASCEND (A Spontaneous Chinese-English Dataset) is a high-quality Mandarin Chinese-English code-switching corpus built on spontaneous multi-turn conversational dialogue sources collected in Hong Kong. We report ASCEND’s design and procedure for collecting the speech data, including annotations. ASCEND consists of 10.62 hours of clean speech, collected from 23 bilingual speakers of Chinese and English. Furthermore, we conduct baseline experiments using pre-trained wav2vec 2.0 models, achieving a best performance of 22.69% character error rate and 27.05% mixed error rate.

Cross-domain named entity recognition (NER) models are able to cope with the scarcity issue of NER samples in target domains. However, most of the existing NER benchmarks lack domain-specialized entity types or do not focus on a certain domain, leading to a less effective cross-domain evaluation. To address these obstacles, we introduce a cross-domain NER dataset (CrossNER), a fully-labeled collection of NER data spanning over five diverse domains with specialized entity categories for different domains. Additionally, we also provide a domain-related corpus since using it to continue pre-training language models (domain-adaptive pre-training) is effective for the domain adaptation. We then conduct comprehensive experiments to explore the effectiveness of leveraging different levels of the domain corpus and pre-training strategies to do domain-adaptive pre-training for the cross-domain task. Results show that focusing on the fractional corpus containing domain-specialized entities and utilizing a more challenging pre-training strategy in domain-adaptive pre-training are beneficial for the NER domain adaptation, and our proposed method can consistently outperform existing cross-domain NER baselines. Nevertheless, experiments also illustrate the challenge of this cross-domain NER task. We hope that our dataset and baselines will catalyze research in the NER domain adaptation area. The code and data are available at https://github.com/zliucr/CrossNER.

Cross-lingual models trained on source language tasks possess the capability to directly transfer to target languages. However, since word order variances generally exist in different languages, cross-lingual models that overfit into the word order of the source language could have sub-optimal performance in target languages. In this paper, we hypothesize that reducing the word order information fitted into the models can improve the adaptation performance in target languages. To verify this hypothesis, we introduce several methods to make models encode less word order information of the source language and test them based on cross-lingual word embeddings and the pre-trained multilingual model. Experimental results on three sequence labeling tasks (i.e., part-of-speech tagging, named entity recognition and slot filling tasks) show that reducing word order information injected into the model can achieve better zero-shot cross-lingual performance. Further analysis illustrates that fitting excessive or insufficient word order information into the model results in inferior cross-lingual performance. Moreover, our proposed methods can also be applied to strong cross-lingual models and further improve their performance.

Multimodal abstractive summarization (MAS) models that summarize videos (vision modality) and their corresponding transcripts (text modality) are able to extract the essential information from massive multimodal data on the Internet.Recently, large-scale generative pretrained language models (GPLMs) have been shown to be effective in text generation tasks.However, existing MAS models cannot leverage GPLMs' powerful generation ability.To fill this research gap, we aim to study two research questions: 1) how to inject visual information into GPLMs without hurting their generation ability; and 2) where is the optimal place in GPLMs to inject the visual information?In this paper, we present a simple yet effective method to construct vision guided (VG) GPLMs for the MAS task using attention-based add-on layers to incorporate visual information while maintaining their original text generation ability.Results show that our best model significantly surpasses the prior state-of-the-art model by 5.7 ROUGE-1, 5.3 ROUGE-2, and 5.1 ROUGE-L scores on the How2 dataset (Sanabria et al., 2018), and our visual guidance method contributes 83.6% of the overall improvement.Furthermore, we conduct thorough ablation studies to analyze the effectiveness of various modality fusion methods and fusion locations.* * The two authors contribute equally.The code is available at: https://github.com/HLTCHKUST/VG-GPLMs Video Frames Transcript: so now we are going to go over some basics sheet music readings for the key of g flat major.so you noticed the key of g flat, when you are reading real books, there is going to be a treble cleft here.it is going to have 6 flats 1, b flat, e flat, a flat, d flat, g flat and c flat.so 6 flats equals key of g flat.[...] so if you have a flat and there is a natural sign, play the a. so go through the scale and you've got g flat, a flat, d flat, c, flat, d flat, e flat and f, so f is your only 9 flat note in the scale.(No mention of the piano) Reference Summary: learn how to read and write music intervals for improving your playing and improvisational skills on the piano in this free video clip series.Summary from Transcript (BART): learn tips on how to read and write intervals on sheet music in this free video clip on music theory and music lessons.Summary from Transcript+Video (VG-BART): learn how to sight read in the key of g flat for improving your playing and improvisational skills on the piano in this free video clip series.

Wenliang Dai, Samuel Cahyawijaya, Zihan Liu, Pascale Fung. Proceedings of the 2021 Conference of the North American Chapter of the Association for Computational Linguistics: Human Language Technologies. 2021.

State-of-the-art abstractive summarization models generally rely on extensive labeled data, which lowers their generalization ability on domains where such data are not available.In this paper, we present a study of domain adaptation for the abstractive summarization task across six diverse target domains in a low-resource setting.Specifically, we investigate the second phase of pre-training on large-scale generative models under three different settings: 1) source domain pre-training; 2) domain-adaptive pre-training; and 3) taskadaptive pre-training.Experiments show that the effectiveness of pre-training is correlated with the similarity between the pre-training data and the target domain task.Moreover, we find that continuing pre-training could lead to the pre-trained model's catastrophic forgetting, and a learning method with less forgetting can alleviate this issue.Furthermore, results illustrate that a huge gap still exists between the low-resource and high-resource settings, which highlights the need for more advanced domain adaptation methods for the abstractive summarization task. 1

We present CAiRE, an end-to-end generative empathetic chatbot designed to recognize user emotions and respond in an empathetic manner. Our system adapts the Generative Pre-trained Transformer (GPT) to empathetic response generation task via transfer learning. CAiRE is built primarily to focus on empathy integration in fully data-driven generative dialogue systems. We create a web-based user interface which allows multiple users to asynchronously chat with CAiRE. CAiRE also collects user feedback and continues to improve its response quality by discarding undesirable generations via active learning and negative training.

Recently, data-driven task-oriented dialogue systems have achieved promising performance in English. However, developing dialogue systems that support low-resource languages remains a long-standing challenge due to the absence of high-quality data. In order to circumvent the expensive and time-consuming data collection, we introduce Attention-Informed Mixed-Language Training (MLT), a novel zero-shot adaptation method for cross-lingual task-oriented dialogue systems. It leverages very few task-related parallel word pairs to generate code-switching sentences for learning the inter-lingual semantics across languages. Instead of manually selecting the word pairs, we propose to extract source words based on the scores computed by the attention layer of a trained English task-related model and then generate word pairs using existing bilingual dictionaries. Furthermore, intensive experiments with different cross-lingual embeddings demonstrate the effectiveness of our approach. Finally, with very few word pairs, our model achieves significant zero-shot adaptation performance improvements in both cross-lingual dialogue state tracking and natural language understanding (i.e., intent detection and slot filling) tasks compared to the current state-of-the-art approaches, which utilize a much larger amount of bilingual data.

As an essential task in task-oriented dialog systems, slot filling requires extensive training data in a certain domain.However, such data are not always available.Hence, cross-domain slot filling has naturally arisen to cope with this data scarcity problem.In this paper, we propose a Coarse-to-fine approach (Coach) for cross-domain slot filling.Our model first learns the general pattern of slot entities by detecting whether the tokens are slot entities or not.It then predicts the specific types for the slot entities.In addition, we propose a template regularization approach to improve the adaptation robustness by regularizing the representation of utterances based on utterance templates.Experimental results show that our model significantly outperforms state-of-theart approaches in slot filling.Furthermore, our model can also be applied to the cross-domain named entity recognition task, and it achieves better adaptation performance than other existing baselines.The code is available at https: //github.com/zliucr/coach.

An increasing number of people in the world today speak a mixed-language as a result of being multilingual.However, building a speech recognition system for code-switching remains difficult due to the availability of limited resources and the expense and significant effort required to collect mixed-language data.We therefore propose a new learning method, meta-transfer learning, to transfer learn on a code-switched speech recognition system in a low-resource setting by judiciously extracting information from high-resource monolingual datasets.Our model learns to recognize individual languages, and transfer them so as to better recognize mixed-language speech by conditioning the optimization on the codeswitching data.Based on experimental results, our model outperforms existing baselines on speech recognition and language modeling tasks, and is faster to converge.

Despite the promising results of current crosslingual models for spoken language understanding systems, they still suffer from imperfect cross-lingual representation alignments between the source and target languages, which makes the performance sub-optimal.To cope with this issue, we propose a regularization approach to further align word-level and sentence-level representations across languages without any external resource.First, we regularize the representation of user utterances based on their corresponding labels.Second, we regularize the latent variable model (Liu et al., 2019a) by leveraging adversarial training to disentangle the latent variables.Experiments on the cross-lingual spoken language understanding task show that our model outperforms current state-of-the-art methods in both few-shot and zero-shot scenarios, and our model, trained on a few-shot setting with only 3% of the target language training data, achieves comparable performance to the supervised training with all the training data. 1

Highly performing deep neural networks come at the cost of computational complexity that limits their practicality for deployment on portable devices. We propose the low-rank transformer (LRT), a memory-efficient and fast neural architecture that significantly reduces the parameters and boosts the speed of training and inference for end-to-end speech recognition. Our approach reduces the number of parameters of the network by more than 50% and speeds up the inference time by around 1.35x compared to the baseline transformer model. The experiments show that our LRT model generalizes better and yields lower error rates on both validation and test sets compared to an uncompressed transformer model. The LRT model outperforms those from existing works on several datasets in an end-to-end setting without using an external language model or acoustic data.

Local dialects influence people to pronounce words of the same language differently from each other. The great variability and complex characteristics of accents create a major challenge for training a robust and accent-agnostic automatic speech recognition (ASR) system. In this paper, we introduce a cross-accented English speech recognition task as a benchmark for measuring the ability of the model to adapt to unseen accents using the existing CommonVoice corpus. We also propose an accent-agnostic approach that extends the model-agnostic meta-learning (MAML) algorithm for fast adaptation to unseen accents. Our approach significantly outperforms joint training in both zero-shot, few-shot, and all-shot in the mixed-region and cross-region settings in terms of word error rate. Copyright © 2020 ISCA

Zihan Liu, Jamin Shin, Yan Xu, Genta Indra Winata, Peng Xu, Andrea Madotto, Pascale Fung. Proceedings of the 2019 Conference on Empirical Methods in Natural Language Processing and the 9th International Joint Conference on Natural Language Processing (EMNLP-IJCNLP). 2019.

Genta Indra Winata, Zhaojiang Lin, Jamin Shin, Zihan Liu, Pascale Fung. Proceedings of the 2019 Conference on Empirical Methods in Natural Language Processing and the 9th International Joint Conference on Natural Language Processing (EMNLP-IJCNLP). 2019.