Large language models (LLMs) have shown impressive few-shot generalization on many tasks via in-context learning (ICL). Despite their success in showing such emergent abilities, the scale and complexity of larger models also lead to unprecedentedly high computational demands and deployment challenges. In reaction, researchers explore transferring the powerful capabilities of larger models to more efficient and compact models by typically aligning the output of smaller models with that of larger models. Existing methods either train smaller models on the generated outputs of larger models or to imitate their token-level probability distributions. However, these distillation methods pay little to no attention to the input part, which also plays a crucial role in ICL. Based on the finding that the performance of ICL is highly sensitive to the selection of demonstration examples, we propose Bidirectional Alignment (BiAlign) to fully leverage the models' preferences for ICL examples to improve the ICL abilities of smaller models. Specifically, we introduce the alignment of input preferences between smaller and larger models by incorporating a novel ranking loss, in addition to aligning the token-level output distribution. With extensive experiments and analysis, we demonstrate that BiAlign can consistently outperform existing baselines on a variety of tasks including language understanding, reasoning, and coding.

Measuring the coherence of text is a vital aspect of evaluating the quality of written content. Recent advancements in neural coherence modeling have demonstrated their efficacy in capturing entity coreference and discourse relations, thereby enhancing coherence evaluation. However, many existing methods heavily depend on static embeddings or focus narrowly on nearby context, constraining their capacity to measure the overarching coherence of long texts. In this paper, we posit that coherent texts inherently manifest a sequential and cohesive interplay among sentences, effectively conveying the central theme, purpose, or standpoint. To explore this abstract relationship, we introduce the "BBScore," a novel reference-free metric grounded in Brownian bridge theory for assessing text coherence. Our findings showcase that when synergized with a simple additional classification component, this metric attains a performance level comparable to state-of-the-art techniques on standard artificial discrimination tasks. We also establish in downstream tasks that this metric effectively differentiates between human-written documents and text generated by large language models under a specific domain. Furthermore, we illustrate the efficacy of this approach in detecting written styles attributed to diverse large language models, underscoring its potential for generalizability. In summary, we present a novel Brownian bridge coherence metric capable of measuring both local and global text coherence, while circumventing the need for end-to-end model training. This flexibility allows for its application in various downstream tasks.

Semi-supervised learning holds great promise for many real-world applications, due to its ability to leverage both unlabeled and expensive labeled data. However, most semi-supervised learning algorithms still heavily rely on the limited labeled data to infer and utilize the hidden information from unlabeled data. We note that any semi-supervised learning task under the self-training paradigm also hides an auxiliary task of discriminating label observability. Jointly solving these two tasks allows full utilization of information from both labeled and unlabeled data, thus alleviating the problem of over-reliance on labeled data. This naturally leads to a new generic and efficient learning framework without the reliance on any domain-specific information, which we call FlexSSL. The key idea of FlexSSL is to construct a semi-cooperative "game", which forges cooperation between a main self-interested semi-supervised learning task and a companion task that infers label observability to facilitate main task training. We show with theoretical derivation of its connection to loss re-weighting on noisy labels. Through evaluations on a diverse range of tasks, we demonstrate that FlexSSL can consistently enhance the performance of semi-supervised learning algorithms.

In the era of advanced multimodel learning, multimodal large language models (MLLMs) such as GPT-4V have made remarkable strides towards bridging language and visual elements. However, the closed-source nature and considerable computational demand present notable challenges for universal usage and modifications. This is where open-source MLLMs like LLaVA and MiniGPT-4 come in, presenting groundbreaking achievements across tasks. Despite these accomplishments, computational efficiency remains an unresolved issue, as these models, like LLaVA-v1.5-13B, require substantial resources. Addressing these issues, we introduce TinyGPT-V, a new-wave model marrying impressive performance with commonplace computational capacity. It stands out by requiring merely a 24G GPU for training and an 8G GPU or CPU for inference. Built upon Phi-2, TinyGPT-V couples an effective language backbone with pre-trained vision modules from BLIP-2 or CLIP. TinyGPT-V's 2.8B parameters can undergo a unique quantisation process, suitable for local deployment and inference tasks on 8G various devices. Our work fosters further developments for designing cost-effective, efficient, and high-performing MLLMs, expanding their applicability in a broad array of real-world scenarios. Furthermore this paper proposed a new paradigm of Multimodal Large Language Model via small backbones. Our code and training weights are placed at: https://github.com/DLYuanGod/TinyGPT-V and https://huggingface.co/Tyrannosaurus/TinyGPT-V respectively.

Previous research has shown that Artificial Intelligence is capable of distinguishing between authentic paintings by a given artist and human-made forgeries with remarkable accuracy, provided sufficient training. However, with the limited amount of existing known forgeries, augmentation methods for forgery detection are highly desirable. In this work, we examine the potential of incorporating synthetic artworks into training datasets to enhance the performance of forgery detection. Our investigation focuses on paintings by Vincent van Gogh, for which we release the first dataset specialized for forgery detection. To reinforce our results, we conduct the same analyses on the artists Amedeo Modigliani and Raphael. We train a classifier to distinguish original artworks from forgeries. For this, we use human-made forgeries and imitations in the style of well-known artists and augment our training sets with images in a similar style generated by Stable Diffusion and StyleGAN. We find that the additional synthetic forgeries consistently improve the detection of human-made forgeries. In addition, we find that, in line with previous research, the inclusion of synthetic forgeries in the training also enables the detection of AI-generated forgeries, especially if created using a similar generator.

While WangchanBERTa has become the de facto standard in transformer-based Thai language modeling, it still has shortcomings in regard to the understanding of foreign words, most notably English words, which are often borrowed without orthographic assimilation into Thai in many contexts. We identify the lack of foreign vocabulary in WangchanBERTa's tokenizer as the main source of these shortcomings. We then expand WangchanBERTa's vocabulary via vocabulary transfer from XLM-R's pretrained tokenizer and pretrain a new model using the expanded tokenizer, starting from WangchanBERTa's checkpoint, on a new dataset that is larger than the one used to train WangchanBERTa. Our results show that our new pretrained model, PhayaThaiBERT, outperforms WangchanBERTa in many downstream tasks and datasets.

Reinforcement learning (RL) has proven to be highly effective in tackling complex decision-making and control tasks. However, prevalent model-free RL methods often face severe performance degradation due to the well-known overestimation issue. In response to this problem, we recently introduced an off-policy RL algorithm, called distributional soft actor-critic (DSAC or DSAC-v1), which can effectively improve the value estimation accuracy by learning a continuous Gaussian value distribution. Nonetheless, standard DSAC has its own shortcomings, including occasionally unstable learning processes and the necessity for task-specific reward scaling, which may hinder its overall performance and adaptability in some special tasks. This paper further introduces three important refinements to standard DSAC in order to address these shortcomings. These refinements consist of expected value substituting, twin value distribution learning, and variance-based critic gradient adjusting. The modified RL algorithm is named as DSAC with three refinements (DSAC-T or DSAC-v2), and its performances are systematically evaluated on a diverse set of benchmark tasks. Without any task-specific hyperparameter tuning, DSAC-T surpasses or matches a lot of mainstream model-free RL algorithms, including SAC, TD3, DDPG, TRPO, and PPO, in all tested environments. Additionally, DSAC-T, unlike its standard version, ensures a highly stable learning process and delivers similar performance across varying reward scales.

Autistic Spectrum Disorder (ASD) is a neurological disease characterized by difficulties with social interaction, communication, and repetitive activities. While its primary origin lies in genetics, early detection is crucial, and leveraging machine learning offers a promising avenue for a faster and more cost-effective diagnosis. This study employs diverse machine learning methods to identify crucial ASD traits, aiming to enhance and automate the diagnostic process. We study eight state-of-the-art classification models to determine their effectiveness in ASD detection. We evaluate the models using accuracy, precision, recall, specificity, F1-score, area under the curve (AUC), kappa, and log loss metrics to find the best classifier for these binary datasets. Among all the classification models, for the children dataset, the SVM and LR models achieve the highest accuracy of 100% and for the adult dataset, the LR model produces the highest accuracy of 97.14%. Our proposed ANN model provides the highest accuracy of 94.24% for the new combined dataset when hyperparameters are precisely tuned for each model. As almost all classification models achieve high accuracy which utilize true labels, we become interested in delving into five popular clustering algorithms to understand model behavior in scenarios without true labels. We calculate Normalized Mutual Information (NMI), Adjusted Rand Index (ARI), and Silhouette Coefficient (SC) metrics to select the best clustering models. Our evaluation finds that spectral clustering outperforms all other benchmarking clustering models in terms of NMI and ARI metrics while demonstrating comparability to the optimal SC achieved by k-means. The implemented code is available at GitHub.

Recently, Large Language Models (LLMs) have shown impressive abilities in code generation. However, existing LLMs' decoding strategies are designed for Natural Language (NL) generation, overlooking the differences between NL and programming languages (PL). Due to this oversight, a better decoding strategy for code generation remains an open question. In this paper, we conduct the first systematic study to explore a decoding strategy specialized in code generation. With an analysis of loss distributions of code tokens, we find that code tokens can be divided into two categories: challenging tokens that are difficult to predict and confident tokens that can be easily inferred. Among them, the challenging tokens mainly appear at the beginning of a code block. Inspired by the above findings, we propose a simple yet effective method: Adaptive Temperature (AdapT) sampling, which dynamically adjusts the temperature coefficient when decoding different tokens. We apply a larger temperature when sampling for challenging tokens, allowing LLMs to explore diverse choices. We employ a smaller temperature for confident tokens avoiding the influence of tail randomness noises. We apply AdapT sampling to LLMs with different sizes and conduct evaluations on two popular datasets. Results show that AdapT sampling significantly outperforms state-of-the-art decoding strategy.

The longstanding goal of multi-lingual learning has been to develop a universal cross-lingual model that can withstand the changes in multilingual data distributions. There has been a large amount of work to adapt such multilingual models to unseen target languages. However, the majority of work in this direction focuses on the standard one-hop transfer learning pipeline from source to target languages, Figure 1: An overview of CCL: We use an example whereas in realistic scenarios, new languages of a non-stationary datastream moving from high to can be incorporated at any time in a sequential low resource languages. Each bold and dashed box manner. In this paper, we present a principled represents either a training or test data instance being Cross-lingual Continual Learning (CCL) evaluation fine-tuned or evaluated on, respectively. To support this paradigm, where we analyze different categories problem setup, we evaluate the cross-lingual capabilities of approaches used to continually adapt of continual approaches. Those capabilities include to emerging data from different languages. We knowledge preservation on old languages, accumulation provide insights into what makes multilingual to the current language, and generalization to sequential learning particularly challenging.