Efficient Many-Shot In-Context Learning with Dynamic Block-Sparse Attention

Many-shot in-context learning has recently shown promise as an alternative to finetuning, with the major advantage that the same model can be served for multiple tasks. However, this shifts the computational burden from training-time to inference-time, making deployment of many-shot ICL challenging to justify in-practice. This cost is further increased if a custom demonstration set is retrieved for each inference example. We present Dynamic Block-Sparse Attention, a training-free framework for retrieval-based many-shot in-context learning. By combining carefully designed block-sparse attention and retrieval of cached groups of demonstrations, we achieve comparable per-example latency to finetuning while maintaining on average >95% of the best method's accuracy across strong ICL and finetuning baselines. We hope that this will further enable the deployment of many-shot ICL at scale.

UIFormer: A Unified Transformer-based Framework for Incremental Few-Shot Object Detection and Instance Segmentation

This paper introduces a novel framework for unified incremental few-shot object detection (iFSOD) and instance segmentation (iFSIS) using the Transformer architecture. Our goal is to create an optimal solution for situations where only a few examples of novel object classes are available, with no access to training data for base or old classes, while maintaining high performance across both base and novel classes. To achieve this, We extend Mask-DINO into a two-stage incremental learning framework. Stage 1 focuses on optimizing the model using the base dataset, while Stage 2 involves fine-tuning the model on novel classes. Besides, we incorporate a classifier selection strategy that assigns appropriate classifiers to the encoder and decoder according to their distinct functions. Empirical evidence indicates that this approach effectively mitigates the over-fitting on novel classes learning. Furthermore, we implement knowledge distillation to prevent catastrophic forgetting of base classes. Comprehensive evaluations on the COCO and LVIS datasets for both iFSIS and iFSOD tasks demonstrate that our method significantly outperforms state-of-the-art approaches.

Seed-Music: A Unified Framework for High Quality and Controlled Music Generation

We introduce Seed-Music, a suite of music generation systems capable of producing high-quality music with fine-grained style control. Our unified framework leverages both auto-regressive language modeling and diffusion approaches to support two key music creation workflows: \textit{controlled music generation} and \textit{post-production editing}. For controlled music generation, our system enables vocal music generation with performance controls from multi-modal inputs, including style descriptions, audio references, musical scores, and voice prompts. For post-production editing, it offers interactive tools for editing lyrics and vocal melodies directly in the generated audio. We encourage readers to listen to demo audio examples at https://team.doubao.com/seed-music .

Effect of Adaptation Rate and Cost Display in a Human-AI Interaction Game

As interactions between humans and AI become more prevalent, it is critical to have better predictors of human behavior in these interactions. We investigated how changes in the AI's adaptive algorithm impact behavior predictions in two-player continuous games. In our experiments, the AI adapted its actions using a gradient descent algorithm under different adaptation rates while human participants were provided cost feedback. The cost feedback was provided by one of two types of visual displays: (a) cost at the current joint action vector, or (b) cost in a local neighborhood of the current joint action vector. Our results demonstrate that AI adaptation rate can significantly affect human behavior, having the ability to shift the outcome between two game theoretic equilibrium. We observed that slow adaptation rates shift the outcome towards the Nash equilibrium, while fast rates shift the outcome towards the human-led Stackelberg equilibrium. The addition of localized cost information had the effect of shifting outcomes towards Nash, compared to the outcomes from cost information at only the current joint action vector. Future work will investigate other effects that influence the convergence of gradient descent games.

MV-DETR: Multi-modality indoor object detection by Multi-View DEtecton TRansformers

We introduce a novel MV-DETR pipeline which is effective while efficient transformer based detection method. Given input RGBD data, we notice that there are super strong pretraining weights for RGB data while less effective works for depth related data. First and foremost , we argue that geometry and texture cues are both of vital importance while could be encoded separately. Secondly, we find that visual texture feature is relatively hard to extract compared with geometry feature in 3d space. Unfortunately, single RGBD dataset with thousands of data is not enough for training an discriminating filter for visual texture feature extraction. Last but certainly not the least, we designed a lightweight VG module consists of a visual textual encoder, a geometry encoder and a VG connector. Compared with previous state of the art works like V-DETR, gains from pretrained visual encoder could be seen. Extensive experiments on ScanNetV2 dataset shows the effectiveness of our method. It is worth mentioned that our method achieve 78\% AP which create new state of the art on ScanNetv2 benchmark.

The Llama 3 Herd of Models

Modern artificial intelligence (AI) systems are powered by foundation models. This paper presents a new set of foundation models, called Llama 3. It is a herd of language models that natively support multilinguality, coding, reasoning, and tool usage. Our largest model is a dense Transformer with 405B parameters and a context window of up to 128K tokens. This paper presents an extensive empirical evaluation of Llama 3. We find that Llama 3 delivers comparable quality to leading language models such as GPT-4 on a plethora of tasks. We publicly release Llama 3, including pre-trained and post-trained versions of the 405B parameter language model and our Llama Guard 3 model for input and output safety. The paper also presents the results of experiments in which we integrate image, video, and speech capabilities into Llama 3 via a compositional approach. We observe this approach performs competitively with the state-of-the-art on image, video, and speech recognition tasks. The resulting models are not yet being broadly released as they are still under development.

A Two-Stage Adverse Weather Semantic Segmentation Method for WeatherProof Challenge CVPR 2024 Workshop UG2+

This technical report presents our team's solution for the WeatherProof Dataset Challenge: Semantic Segmentation in Adverse Weather at CVPR'24 UG2+. We propose a two-stage deep learning framework for this task. In the first stage, we preprocess the provided dataset by concatenating images into video sequences. Subsequently, we leverage a low-rank video deraining method to generate high-fidelity pseudo ground truths. These pseudo ground truths offer superior alignment compared to the original ground truths, facilitating model convergence during training. In the second stage, we employ the InternImage network to train for the semantic segmentation task using the generated pseudo ground truths. Notably, our meticulously designed framework demonstrates robustness to degraded data captured under adverse weather conditions. In the challenge, our solution achieved a competitive score of 0.43 on the Mean Intersection over Union (mIoU) metric, securing a respectable rank of 4th.

Trusted Multi-view Learning with Label Noise

Multi-view learning methods often focus on improving decision accuracy while neglecting the decision uncertainty, which significantly restricts their applications in safety-critical applications. To address this issue, researchers propose trusted multi-view methods that learn the class distribution for each instance, enabling the estimation of classification probabilities and uncertainty. However, these methods heavily rely on high-quality ground-truth labels. This motivates us to delve into a new generalized trusted multi-view learning problem: how to develop a reliable multi-view learning model under the guidance of noisy labels? We propose a trusted multi-view noise refining method to solve this problem. We first construct view-opinions using evidential deep neural networks, which consist of belief mass vectors and uncertainty estimates. Subsequently, we design view-specific noise correlation matrices that transform the original opinions into noisy opinions aligned with the noisy labels. Considering label noises originating from low-quality data features and easily-confused classes, we ensure that the diagonal elements of these matrices are inversely proportional to the uncertainty, while incorporating class relations into the off-diagonal elements. Finally, we aggregate the noisy opinions and employ a generalized maximum likelihood loss on the aggregated opinion for model training, guided by the noisy labels. We empirically compare TMNR with state-of-the-art trusted multi-view learning and label noise learning baselines on 5 publicly available datasets. Experiment results show that TMNR outperforms baseline methods on accuracy, reliability and robustness. We promise to release the code and all datasets on Github and show the link here.

HO-Gaussian: Hybrid Optimization of 3D Gaussian Splatting for Urban Scenes

The rapid growth of 3D Gaussian Splatting (3DGS) has revolutionized neural rendering, enabling real-time production of high-quality renderings. However, the previous 3DGS-based methods have limitations in urban scenes due to reliance on initial Structure-from-Motion(SfM) points and difficulties in rendering distant, sky and low-texture areas. To overcome these challenges, we propose a hybrid optimization method named HO-Gaussian, which combines a grid-based volume with the 3DGS pipeline. HO-Gaussian eliminates the dependency on SfM point initialization, allowing for rendering of urban scenes, and incorporates the Point Densitification to enhance rendering quality in problematic regions during training. Furthermore, we introduce Gaussian Direction Encoding as an alternative for spherical harmonics in the rendering pipeline, which enables view-dependent color representation. To account for multi-camera systems, we introduce neural warping to enhance object consistency across different cameras. Experimental results on widely used autonomous driving datasets demonstrate that HO-Gaussian achieves photo-realistic rendering in real-time on multi-camera urban datasets.

VIALM: A Survey and Benchmark of Visually Impaired Assistance with Large Models

Visually Impaired Assistance (VIA) aims to automatically help visually impaired (VI) handle daily activities. The advancement of VIA primarily depends on developments in Computer Vision (CV) and Natural Language Processing (NLP), both of which exhibit cutting-edge paradigms with large models (LMs). Furthermore, LMs have shown exceptional multimodal abilities to tackle challenging physically-grounded tasks such as embodied robots. To investigate the potential and limitations of state-of-the-art (SOTA) LMs' capabilities in VIA applications, we present an extensive study for the task of VIA with LMs (\textbf{VIALM}). In this task, given an \textit{image} illustrating the physical environments and a \textit{linguistic request} from a VI user, VIALM aims to output step-by-step \textit{guidance} to assist the VI user in fulfilling the request grounded in the environment. The study consists of a survey reviewing recent LM research and benchmark experiments examining selected LMs' capabilities in VIA. The results indicate that while LMs can augment VIA, their output cannot be well \textit{grounded} in reality (i.e., 25.7\% GPT-4's responses) and lacks \textit{fine-grained} guidance (i.e., 32.1\% GPT-4's responses).