FoPru: Focal Pruning for Efficient Large Vision-Language Models

Large Vision-Language Models (LVLMs) represent a significant advancement toward achieving superior multimodal capabilities by enabling powerful Large Language Models (LLMs) to understand visual input. Typically, LVLMs utilize visual encoders, such as CLIP, to transform images into visual tokens, which are then aligned with textual tokens through projection layers before being input into the LLM for inference. Although existing LVLMs have achieved significant success, their inference efficiency is still limited by the substantial number of visual tokens and the potential redundancy among them. To mitigate this issue, we propose Focal Pruning (FoPru), a training-free method that prunes visual tokens based on the attention-based token significance derived from the vision encoder. Specifically, we introduce two alternative pruning strategies: 1) the rank strategy, which leverages all token significance scores to retain more critical tokens in a global view; 2) the row strategy, which focuses on preserving continuous key information in images from a local perspective. Finally, the selected tokens are reordered to maintain their original positional relationships. Extensive experiments across various LVLMs and multimodal datasets demonstrate that our method can prune a large number of redundant tokens while maintaining high accuracy, leading to significant improvements in inference efficiency.

Chanel-Orderer: A Channel-Ordering Predictor for Tri-Channel Natural Images

This paper shows a proof-of-concept that, given a typical 3-channel images but in a randomly permuted channel order, a model (termed as Chanel-Orderer) with ad-hoc inductive biases in terms of both architecture and loss functions can accurately predict the channel ordering and knows how to make it right. Specifically, Chanel-Orderer learns to score each of the three channels with the priors of object semantics and uses the resulting scores to predict the channel ordering. This brings up benefits into a typical scenario where an \texttt{RGB} image is often mis-displayed in the \texttt{BGR} format and needs to be corrected into the right order. Furthermore, as a byproduct, the resulting model Chanel-Orderer is able to tell whether a given image is a near-gray-scale image (near-monochromatic) or not (polychromatic). Our research suggests that Chanel-Orderer mimics human visual coloring of our physical natural world.

Hunyuan-Large: An Open-Source MoE Model with 52 Billion Activated Parameters by Tencent

In this paper, we introduce Hunyuan-Large, which is currently the largest open-source Transformer-based mixture of experts model, with a total of 389 billion parameters and 52 billion activation parameters, capable of handling up to 256K tokens. We conduct a thorough evaluation of Hunyuan-Large's superior performance across various benchmarks including language understanding and generation, logical reasoning, mathematical problem-solving, coding, long-context, and aggregated tasks, where it outperforms LLama3.1-70B and exhibits comparable performance when compared to the significantly larger LLama3.1-405B model. Key practice of Hunyuan-Large include large-scale synthetic data that is orders larger than in previous literature, a mixed expert routing strategy, a key-value cache compression technique, and an expert-specific learning rate strategy. Additionally, we also investigate the scaling laws and learning rate schedule of mixture of experts models, providing valuable insights and guidances for future model development and optimization. The code and checkpoints of Hunyuan-Large are released to facilitate future innovations and applications. Codes: https://github.com/Tencent/Hunyuan-Large Models: https://huggingface.co/tencent/Tencent-Hunyuan-Large

LLMCO2: Advancing Accurate Carbon Footprint Prediction for LLM Inferences

Throughout its lifecycle, a large language model (LLM) generates a substantially larger carbon footprint during inference than training. LLM inference requests vary in batch size, prompt length, and token generation number, while cloud providers employ different GPU types and quantities to meet diverse service-level objectives for accuracy and latency. It is crucial for both users and cloud providers to have a tool that quickly and accurately estimates the carbon impact of LLM inferences based on a combination of inference request and hardware configurations before execution. Estimating the carbon footprint of LLM inferences is more complex than training due to lower and highly variable model FLOPS utilization, rendering previous equation-based models inaccurate. Additionally, existing machine learning (ML) prediction methods either lack accuracy or demand extensive training data, as they inadequately handle the distinct prefill and decode phases, overlook hardware-specific features, and inefficiently sample uncommon inference configurations. We introduce \coo, a graph neural network (GNN)-based model that greatly improves the accuracy of LLM inference carbon footprint predictions compared to previous methods.

GroupDebate: Enhancing the Efficiency of Multi-Agent Debate Using Group Discussion

In recent years, Large Language Models (LLMs) have demonstrated remarkable capabilities across diverse NLP tasks. Extensive research has explored how to enhance the logical reasoning abilities such as Chain-of-Thought, Chain-of-Thought with Self-Consistency, Tree-Of-Thoughts, and multi-agent debates. In the context of multi-agent debates, significant performance improvements can be achieved with an increasing number of agents and debate rounds. However, the escalation in the number of agents and debate rounds can drastically raise the tokens cost of debates, thereby limiting the scalability of the multi-agent debate technique. To better harness the advantages of multi-agent debates in logical reasoning tasks, this paper proposes a method to significantly reduce token cost in multi-agent debates. This approach involves dividing all agents into multiple debate groups, with agents engaging in debates within their respective groups and sharing interim debate results between groups. Comparative experiments across multiple datasets have demonstrated that this method can reduce the total tokens by up to 51.7% during debates and while potentially enhancing accuracy by as much as 25%. Our method significantly enhances the performance and efficiency of interactions in the multi-agent debate.

Threshold Filtering Packing for Supervised Fine-Tuning: Training Related Samples within Packs

Packing for Supervised Fine-Tuning (SFT) in autoregressive models involves concatenating data points of varying lengths until reaching the designed maximum length to facilitate GPU processing. However, randomly concatenating data points and feeding them into an autoregressive transformer can lead to cross-contamination of sequences due to the significant difference in their subject matter. The mainstream approaches in SFT ensure that each token in the attention calculation phase only focuses on tokens within its own short sequence, without providing additional learning signals for the preceding context. To address these challenges, we introduce Threshold Filtering Packing (TFP), a method that selects samples with related context while maintaining sufficient diversity within the same pack. Our experiments show that TFP offers a simple-to-implement and scalable approach that significantly enhances SFT performance, with observed improvements of up to 7\% on GSM8K, 4\% on HumanEval, and 15\% on the adult-census-income dataset.

Adaptive Differentially Private Structural Entropy Minimization for Unsupervised Social Event Detection

Social event detection refers to extracting relevant message clusters from social media data streams to represent specific events in the real world. Social event detection is important in numerous areas, such as opinion analysis, social safety, and decision-making. Most current methods are supervised and require access to large amounts of data. These methods need prior knowledge of the events and carry a high risk of leaking sensitive information in the messages, making them less applicable in open-world settings. Therefore, conducting unsupervised detection while fully utilizing the rich information in the messages and protecting data privacy remains a significant challenge. To this end, we propose a novel social event detection framework, ADP-SEMEvent, an unsupervised social event detection method that prioritizes privacy. Specifically, ADP-SEMEvent is divided into two stages, i.e., the construction stage of the private message graph and the clustering stage of the private message graph. In the first stage, an adaptive differential privacy approach is used to construct a private message graph. In this process, our method can adaptively apply differential privacy based on the events occurring each day in an open environment to maximize the use of the privacy budget. In the second stage, to address the reduction in data utility caused by noise, a novel 2-dimensional structural entropy minimization algorithm based on optimal subgraphs is used to detect events in the message graph. The highlight of this process is unsupervised and does not compromise differential privacy. Extensive experiments on two public datasets demonstrate that ADP-SEMEvent can achieve detection performance comparable to state-of-the-art methods while maintaining reasonable privacy budget parameters.

Table-Filling via Mean Teacher for Cross-domain Aspect Sentiment Triplet Extraction

Cross-domain Aspect Sentiment Triplet Extraction (ASTE) aims to extract fine-grained sentiment elements from target domain sentences by leveraging the knowledge acquired from the source domain. Due to the absence of labeled data in the target domain, recent studies tend to rely on pre-trained language models to generate large amounts of synthetic data for training purposes. However, these approaches entail additional computational costs associated with the generation process. Different from them, we discover a striking resemblance between table-filling methods in ASTE and two-stage Object Detection (OD) in computer vision, which inspires us to revisit the cross-domain ASTE task and approach it from an OD standpoint. This allows the model to benefit from the OD extraction paradigm and region-level alignment. Building upon this premise, we propose a novel method named \textbf{T}able-\textbf{F}illing via \textbf{M}ean \textbf{T}eacher (TFMT). Specifically, the table-filling methods encode the sentence into a 2D table to detect word relations, while TFMT treats the table as a feature map and utilizes a region consistency to enhance the quality of those generated pseudo labels. Additionally, considering the existence of the domain gap, a cross-domain consistency based on Maximum Mean Discrepancy is designed to alleviate domain shift problems. Our method achieves state-of-the-art performance with minimal parameters and computational costs, making it a strong baseline for cross-domain ASTE.

Using joint angles based on the international biomechanical standards for human action recognition and related tasks

Keypoint data has received a considerable amount of attention in machine learning for tasks like action detection and recognition. However, human experts in movement such as doctors, physiotherapists, sports scientists and coaches use a notion of joint angles standardised by the International Society of Biomechanics to precisely and efficiently communicate static body poses and movements. In this paper, we introduce the basic biomechanical notions and show how they can be used to convert common keypoint data into joint angles that uniquely describe the given pose and have various desirable mathematical properties, such as independence of both the camera viewpoint and the person performing the action. We experimentally demonstrate that the joint angle representation of keypoint data is suitable for machine learning applications and can in some cases bring an immediate performance gain. The use of joint angles as a human meaningful representation of kinematic data is in particular promising for applications where interpretability and dialog with human experts is important, such as many sports and medical applications. To facilitate further research in this direction, we will release a python package to convert keypoint data into joint angles as outlined in this paper.

GCN-DevLSTM: Path Development for Skeleton-Based Action Recognition

Skeleton-based action recognition (SAR) in videos is an important but challenging task in computer vision. The recent state-of-the-art models for SAR are primarily based on graph convolutional neural networks (GCNs), which are powerful in extracting the spatial information of skeleton data. However, it is yet clear that such GCN-based models can effectively capture the temporal dynamics of human action sequences. To this end, we propose the DevLSTM module, which exploits the path development -- a principled and parsimonious representation for sequential data by leveraging the Lie group structure. The path development, originated from Rough path theory, can effectively capture the order of events in high-dimensional stream data with massive dimension reduction and consequently enhance the LSTM module substantially. Our proposed G-DevLSTM module can be conveniently plugged into the temporal graph, complementing existing advanced GCN-based models. Our empirical studies on the NTU60, NTU120 and Chalearn2013 datasets demonstrate that our proposed hybrid model significantly outperforms the current best-performing methods in SAR tasks. The code is available at https://github.com/DeepIntoStreams/GCN-DevLSTM.