Zebra-Llama: Towards Extremely Efficient Hybrid Models

With the growing demand for deploying large language models (LLMs) across diverse applications, improving their inference efficiency is crucial for sustainable and democratized access. However, retraining LLMs to meet new user-specific requirements is prohibitively expensive and environmentally unsustainable. In this work, we propose a practical and scalable alternative: composing efficient hybrid language models from existing pre-trained models. Our approach, Zebra-Llama, introduces a family of 1B, 3B, and 8B hybrid models by combining State Space Models (SSMs) and Multi-head Latent Attention (MLA) layers, using a refined initialization and post-training pipeline to efficiently transfer knowledge from pre-trained Transformers. Zebra-Llama achieves Transformer-level accuracy with near-SSM efficiency using only 7-11B training tokens (compared to trillions of tokens required for pre-training) and an 8B teacher. Moreover, Zebra-Llama dramatically reduces KV cache size -down to 3.9%, 2%, and 2.73% of the original for the 1B, 3B, and 8B variants, respectively-while preserving 100%, 100%, and >97% of average zero-shot performance on LM Harness tasks. Compared to models like MambaInLLaMA, X-EcoMLA, Minitron, and Llamba, Zebra-Llama consistently delivers competitive or superior accuracy while using significantly fewer tokens, smaller teachers, and vastly reduced KV cache memory. Notably, Zebra-Llama-8B surpasses Minitron-8B in few-shot accuracy by 7% while using 8x fewer training tokens, over 12x smaller KV cache, and a smaller teacher (8B vs. 15B). It also achieves 2.6x-3.8x higher throughput (tokens/s) than MambaInLlama up to a 32k context length. We will release code and model checkpoints upon acceptance.

ELIS: Efficient LLM Iterative Scheduling System with Response Length Predictor

We propose ELIS, a serving system for Large Language Models (LLMs) featuring an Iterative Shortest Remaining Time First (ISRTF) scheduler designed to efficiently manage inference tasks with the shortest remaining tokens. Current LLM serving systems often employ a first-come-first-served scheduling strategy, which can lead to the "head-of-line blocking" problem. To overcome this limitation, it is necessary to predict LLM inference times and apply a shortest job first scheduling strategy. However, due to the auto-regressive nature of LLMs, predicting the inference latency is challenging. ELIS addresses this challenge by training a response length predictor for LLMs using the BGE model, an encoder-based state-of-the-art model. Additionally, we have devised the ISRTF scheduling strategy, an optimization of shortest remaining time first tailored to existing LLM iteration batching. To evaluate our work in an industrial setting, we simulate streams of requests based on our study of real-world user LLM serving trace records. Furthermore, we implemented ELIS as a cloud-native scheduler system on Kubernetes to evaluate its performance in production environments. Our experimental results demonstrate that ISRTF reduces the average job completion time by up to 19.6%.

X-EcoMLA: Upcycling Pre-Trained Attention into MLA for Efficient and Extreme KV Compression

Multi-head latent attention (MLA) is designed to optimize KV cache memory through low-rank key-value joint compression. Rather than caching keys and values separately, MLA stores their compressed latent representations, reducing memory overhead while maintaining the performance. While MLA improves memory efficiency without compromising language model accuracy, its major limitation lies in its integration during the pre-training phase, requiring models to be trained from scratch. This raises a key question: can we use MLA's benefits fully or partially in models that have already been pre-trained with different attention mechanisms? In this paper, we propose X-EcoMLA to deploy post training distillation to enable the upcycling of Transformer-based attention into an efficient hybrid (i.e., combination of regular attention and MLA layers) or full MLA variant through lightweight post-training adaptation, bypassing the need for extensive pre-training. We demonstrate that leveraging the dark knowledge of a well-trained model can enhance training accuracy and enable extreme KV cache compression in MLA without compromising model performance. Our results show that using an 8B teacher model allows us to compress the KV cache size of the Llama3.2-1B-Inst baseline by 6.4x while preserving 100% of its average score across multiple tasks on the LM Harness Evaluation benchmark. This is achieved with only 3.6B training tokens and about 70 GPU hours on AMD MI300 GPUs, compared to the 370K GPU hours required for pre-training the Llama3.2-1B model.

SAM-guided Pseudo Label Enhancement for Multi-modal 3D Semantic Segmentation

Multi-modal 3D semantic segmentation is vital for applications such as autonomous driving and virtual reality (VR). To effectively deploy these models in real-world scenarios, it is essential to employ cross-domain adaptation techniques that bridge the gap between training data and real-world data. Recently, self-training with pseudo-labels has emerged as a predominant method for cross-domain adaptation in multi-modal 3D semantic segmentation. However, generating reliable pseudo-labels necessitates stringent constraints, which often result in sparse pseudo-labels after pruning. This sparsity can potentially hinder performance improvement during the adaptation process. We propose an image-guided pseudo-label enhancement approach that leverages the complementary 2D prior knowledge from the Segment Anything Model (SAM) to introduce more reliable pseudo-labels, thereby boosting domain adaptation performance. Specifically, given a 3D point cloud and the SAM masks from its paired image data, we collect all 3D points covered by each SAM mask that potentially belong to the same object. Then our method refines the pseudo-labels within each SAM mask in two steps. First, we determine the class label for each mask using majority voting and employ various constraints to filter out unreliable mask labels. Next, we introduce Geometry-Aware Progressive Propagation (GAPP) which propagates the mask label to all 3D points within the SAM mask while avoiding outliers caused by 2D-3D misalignment. Experiments conducted across multiple datasets and domain adaptation scenarios demonstrate that our proposed method significantly increases the quantity of high-quality pseudo-labels and enhances the adaptation performance over baseline methods.

Playable Game Generation

In recent years, Artificial Intelligence Generated Content (AIGC) has advanced from text-to-image generation to text-to-video and multimodal video synthesis. However, generating playable games presents significant challenges due to the stringent requirements for real-time interaction, high visual quality, and accurate simulation of game mechanics. Existing approaches often fall short, either lacking real-time capabilities or failing to accurately simulate interactive mechanics. To tackle the playability issue, we propose a novel method called \emph{PlayGen}, which encompasses game data generation, an autoregressive DiT-based diffusion model, and a comprehensive playability-based evaluation framework. Validated on well-known 2D and 3D games, PlayGen achieves real-time interaction, ensures sufficient visual quality, and provides accurate interactive mechanics simulation. Notably, these results are sustained even after over 1000 frames of gameplay on an NVIDIA RTX 2060 GPU. Our code is publicly available: https://github.com/GreatX3/Playable-Game-Generation. Our playable demo generated by AI is: http://124.156.151.207.

A Survey of Attacks on Large Vision-Language Models: Resources, Advances, and Future Trends

With the significant development of large models in recent years, Large Vision-Language Models (LVLMs) have demonstrated remarkable capabilities across a wide range of multimodal understanding and reasoning tasks. Compared to traditional Large Language Models (LLMs), LVLMs present great potential and challenges due to its closer proximity to the multi-resource real-world applications and the complexity of multi-modal processing. However, the vulnerability of LVLMs is relatively underexplored, posing potential security risks in daily usage. In this paper, we provide a comprehensive review of the various forms of existing LVLM attacks. Specifically, we first introduce the background of attacks targeting LVLMs, including the attack preliminary, attack challenges, and attack resources. Then, we systematically review the development of LVLM attack methods, such as adversarial attacks that manipulate model outputs, jailbreak attacks that exploit model vulnerabilities for unauthorized actions, prompt injection attacks that engineer the prompt type and pattern, and data poisoning that affects model training. Finally, we discuss promising research directions in the future. We believe that our survey provides insights into the current landscape of LVLM vulnerabilities, inspiring more researchers to explore and mitigate potential safety issues in LVLM developments. The latest papers on LVLM attacks are continuously collected in https://github.com/liudaizong/Awesome-LVLM-Attack.

Diffusion-Aided Joint Source Channel Coding For High Realism Wireless Image Transmission

Deep learning-based joint source-channel coding (deep JSCC) has been demonstrated as an effective approach for wireless image transmission. Nevertheless, current research has concentrated on minimizing a standard distortion metric such as Mean Squared Error (MSE), which does not necessarily improve the perceptual quality. To address this issue, we propose DiffJSCC, a novel framework that leverages pre-trained text-to-image diffusion models to enhance the realism of images transmitted over the channel. The proposed DiffJSCC utilizes prior deep JSCC frameworks to deliver an initial reconstructed image at the receiver. Then, the spatial and textual features are extracted from the initial reconstruction, which, together with the channel state information (e.g., signal-to-noise ratio, SNR), are passed to a control module to fine-tune the pre-trained Stable Diffusion model. Extensive experiments on the Kodak dataset reveal that our method significantly surpasses both conventional methods and prior deep JSCC approaches on perceptual metrics such as LPIPS and FID scores, especially with poor channel conditions and limited bandwidth. Notably, DiffJSCC can achieve highly realistic reconstructions for 768x512 pixel Kodak images with only 3072 symbols (<0.008 symbols per pixel) under 1dB SNR. Our code will be released in https://github.com/mingyuyng/DiffJSCC.

Diving with Penguins: Detecting Penguins and their Prey in Animal-borne Underwater Videos via Deep Learning

African penguins (Spheniscus demersus) are an endangered species. Little is known regarding their underwater hunting strategies and associated predation success rates, yet this is essential for guiding conservation. Modern bio-logging technology has the potential to provide valuable insights, but manually analysing large amounts of data from animal-borne video recorders (AVRs) is time-consuming. In this paper, we publish an animal-borne underwater video dataset of penguins and introduce a ready-to-deploy deep learning system capable of robustly detecting penguins (mAP50@98.0%) and also instances of fish (mAP50@73.3%). We note that the detectors benefit explicitly from air-bubble learning to improve accuracy. Extending this detector towards a dual-stream behaviour recognition network, we also provide the first results for identifying predation behaviour in penguin underwater videos. Whilst results are promising, further work is required for useful applicability of predation behaviour detection in field scenarios. In summary, we provide a highly reliable underwater penguin detector, a fish detector, and a valuable first attempt towards an automated visual detection of complex behaviours in a marine predator. We publish the networks, the DivingWithPenguins video dataset, annotations, splits, and weights for full reproducibility and immediate usability by practitioners.

Siamese Learning-based Monarch Butterfly Localization

A new GPS-less, daily localization method is proposed with deep learning sensor fusion that uses daylight intensity and temperature sensor data for Monarch butterfly tracking. Prior methods suffer from the location-independent day length during the equinox, resulting in high localization errors around that date. This work proposes a new Siamese learning-based localization model that improves the accuracy and reduces the bias of daily Monarch butterfly localization using light and temperature measurements. To train and test the proposed algorithm, we use $5658$ daily measurement records collected through a data measurement campaign involving 306 volunteers across the U.S., Canada, and Mexico from 2018 to 2020. This model achieves a mean absolute error of $1.416^\circ$ in latitude and $0.393^\circ$ in longitude coordinates outperforming the prior method.

Efficient Computation Sharing for Multi-Task Visual Scene Understanding

Solving multiple visual tasks using individual models can be resource-intensive, while multi-task learning can conserve resources by sharing knowledge across different tasks. Despite the benefits of multi-task learning, such techniques can struggle with balancing the loss for each task, leading to potential performance degradation. We present a novel computation- and parameter-sharing framework that balances efficiency and accuracy to perform multiple visual tasks utilizing individually-trained single-task transformers. Our method is motivated by transfer learning schemes to reduce computational and parameter storage costs while maintaining the desired performance. Our approach involves splitting the tasks into a base task and the other sub-tasks, and sharing a significant portion of activations and parameters/weights between the base and sub-tasks to decrease inter-task redundancies and enhance knowledge sharing. The evaluation conducted on NYUD-v2 and PASCAL-context datasets shows that our method is superior to the state-of-the-art transformer-based multi-task learning techniques with higher accuracy and reduced computational resources. Moreover, our method is extended to video stream inputs, further reducing computational costs by efficiently sharing information across the temporal domain as well as the task domain. Our codes and models will be publicly available.