Efficient High-Resolution Image Synthesis with Linear Diffusion Transformer

Sana is a platform designed for efficient data management and collaboration in machine learning projects, enhancing productivity through streamlined workflows.

Un Ministral, Des Ministraux

Mistral AI has launched Ministral 3B and Ministral 8B, cutting-edge models designed for on-device computing and edge applications, enhancing efficiency and reasoning capabilities in the sub-10B category.

Meta's open AI hardware vision

Meta's latest AI hardware innovations showcased at the OCP Global Summit include the Catalina rack and Grand Teton platform, designed to support advanced AI workloads and enhance collaboration within the open hardware community.

Ichigo: Local real-time voice AI

Ichigo, formerly known as llama3-s, is a local real-time voice AI that enhances text-based LLMs with native listening capabilities, utilizing an early fusion technique inspired by Meta's Chameleon paper.

AI PCs Aren't Good at AI: The CPU Beats the NPU

Qualcomm's NPU on the Microsoft Surface Tablet achieves only 1.3% of its claimed 45 Teraops/s, indicating significant performance gaps compared to expectations and other platforms like Android.

The Future of Big Iron: An Interview with IBM’s Christian Jacobi

Telum II enhances IBM's mainframe capabilities with 8 cores, a 5nm process, and increased L2 SRAM from 256 MB to 360 MB, integrating a DPU for improved I/O performance and scalability.

[R] Switch EMA: A Free Lunch for Better Flatness and Sharpness

Switch EMA (SEMA) enhances Exponential Moving Average (EMA) by modifying parameters post-epoch, leading to improved generalization in deep neural networks (DNNs) without additional costs.

MoH: Multi-Head Attention as Mixture-of-Head Attention

MoH (Mixture-of-Head attention) enhances the multi-head attention mechanism by allowing tokens to select relevant attention heads, improving efficiency while maintaining or exceeding accuracy levels.

MLLM can see? Dynamic Correction Decoding for Hallucination Mitigation

MLLMs can recognize visual objects in earlier layers despite generating incorrect outputs, suggesting that strong knowledge priors may suppress visual information, leading to hallucinations.

Show HN: Automated smooth Nth order derivatives of noisy data

kalmangrad is a Python package that utilizes Bayesian filtering to compute automated smooth N'th order derivatives from non-uniformly sampled time series data, significantly reducing noise impact compared to traditional methods.

The Path to Achieve PyTorch Performance Boost on Windows CPU

PyTorch's CPU performance on Windows has significantly improved with the introduction of mimalloc in version 2.1.2 and SIMD optimizations in version 2.4.1, addressing previous inefficiencies in memory allocation and vectorization.

TwoMinutePapers - 5 Almost Impossible Experiments…Until Now!

[P] Introducing CVPal: A Computer Vision Library for Creating Custom Datasets with Just a Prompt!

CVPal is a groundbreaking computer vision library that enables users to create complete datasets from a simple prompt, now supporting Synthetic Data generation via models like Dalle and Stable Diffusion.

LR-SQL: A Supervised Fine-Tuning Method for Text2SQL Tasks under Low-Resource Scenarios

LR-SQL introduces a novel approach to Text2SQL tasks by utilizing two supervised fine-tuning models, effectively managing database complexity to optimize GPU memory usage during training.

Automatically Generating Visual Hallucination Test Cases for Multimodal Large Language Models

VHExpansion is the first automated method to generate visual hallucination (VH) test cases for multimodal large language models (MLLMs), enhancing the testing process by perturbing questions, answers, and images.